Chapter 7
PAYOUT POLICY
FRANKLIN ALLEN*
University of Pennsylvania
RONI MICHAELY
Cornell University and IDC
Contents
Abstract
Keywords
1 Introduction
2 Some empirical observations on payout policies
3 The Miller-Modigliani dividend irrelevance proposition
4 How should we measure payout?
5 Taxes
5.1
Static models
5.1 1 The role of risk
5.2 Dynamic models
5.2 1 Dynamic tax avoidance strategies
5.2 2 Dynamic ex-dividend day strategies
5.3 Dividends and taxes conclusions
6 Asymmetric information and incomplete contracts
theory
6.1 Signaling and adverse selection models
6.2 Incomplete contracts agency models
7 Empirical evidence
339
339
340
342
351
354
358
359
363
368
368
369
376
377
377
383
8 Transaction costs and other explanations
9 Repurchases
386
386
396
399
404
9.1 The mechanics and some stylized facts
9.2 Theories of repurchases
404
407
9.3 Repurchases compared to dividends
408
7.1 Asymmetric information and signaling models
7.2 Agency models
We are in debt to Gustavo Grullon for his insights and help on this project We would like to thank
Harry DeAngelo, Eric Lie, Ren 6 Stulz and Jeff Wurgler for their comments and suggestions We would
also like to thank Meenakshi Sinha for her research assistance.
Handbook of the Economics of Finance, Edited by G M Constantinides, M Harris and R Stulz
© 2003 Elsevier B V All rights reserved
338
EAllen and R Michaely
9.4 Empirical evidence
9.4 1 How to measure share repurchase activity?
9.4 2 Empirical tests of repurchase theories
9.4 3 Some empirical evidence on dividends compared to share repurchases
9.5 Summary
10 Concluding remarks
References
410
410
412
415
419
420
422
Ch 7:
Payout Policy
339
Abstract
This paper surveys the literature on payout policy We start out by discussing
several stylized facts that are important to the development of any comprehensive
payout policy framework We then describe the Miller and Modigliani ( 1961) payout
irrelevance proposition, and consider the effect of relaxing the assumptions on which
it is based We consider the role of taxes, asymmetric information, incomplete
contracting possibilities, and transaction costs.
The tax-related literature on dividends explores the implications of differential taxes
on dividends and capital gains on stocks' valuation and firms' propensity to pay out
cash in the form of dividends The issues investigated in this literature are of central
importance to corporate finance and asset pricing It is important to understand the
degree to which investor taxes are impounded into security prices, which in turn can
affect investment returns, the cost of capital, capital structure, investment spending, and
governmental revenue collection The overall empirical evidence on this issue appears
to indicate that from a tax perspective, dividends should be minimized.
We review the theoretical as well as empirical literature on Signaling/Adverse
Selection models and Agency models The accumulated evidence indicates that
changes in payout policies are not motivated by firms' desire to signal their true worth
to the market There is no evidence that firms that increase their dividends experience
an unexpectedly high earnings or cash flow in subsequent periods The literature does
point out however, that changes in cash payments are negatively associated with firms'
risk profile This and other evidence seem to be consistent with the notion that both
dividends and repurchases are paid when firms have excess cash flows in order to
reduce potential overinvestment by management.
We also review the issue of the form of payout and the increased tendency to use
open market share repurchases Evidence suggests that the rise in the popularity of
repurchases increases overall payout and increases firms' financial flexibility It seems
that young, risky firms prefer to use repurchases rather then dividends We also observe
that many large, established firms and those with more volatile earnings substitute
repurchases for dividends We believe that the choice of payout method and how payout
policy interacts with capital-structure decisions (such as debt and equity issuance) are
important questions and a promising field for further research.
Keywords
dividends, repurchases, payout policy, asymmetric information, agency problems,
taxes
JEL classification: G 30, G32, G35
340
E Allen and R Michaely
1 Introduction
How much cash should firms give back to their shareholders? And what form should
the payment take? Should corporations pay their shareholders through dividends or
by repurchasing their shares, which is the least costly form of payout from a tax
perspective? Firms must make these important decisions over and over again (some
must be repeated and some need to be reevaluated each period), on a regular basis.
Because these decisions are dynamic they are labeled as payout policy The word
"policy" implies some consistency over time, and that payouts, and dividends in
particular, do not simply evolve in an arbitrary and random manner Much of the
literature in the past forty years has attempted to find and explain the pattern in payout
policies of corporations.
The money involved in these payout decisions is substantial For example, in 1999
corporations spent more than $350b on dividends and repurchases and over $400 b on
liquidating dividends in the form of cash spent on mergers and acquisitions
Payout policy is important not only because of the amount of money involved and the
repeated nature of the decision, but also because payout policy is closely related to, and
interacts with, most of the financial and investment decisions firms make Management
and the board of directors must decide the level of dividends, what repurchases to make
(the mirror image decision of equity issuance), the amount of financial slack the firm
carries (which may be a non-trivial amount; for example, at the end of 1999, Microsoft
held over $17b in financial slack), investment in real assets, mergers and acquisitions,
and debt issuance Since capital markets are neither perfect nor complete, all of these
decisions interact with one another.
Understanding payout policy may also help us to better understand the other pieces
in this puzzle Theories of capital structure, mergers and acquisitions, asset pricing,
and capital budgeting all rely on a view of how and why firms pay out cash.
Six empirical observations play an important role in discussions of payout policies:
(1) Large, established corporations typically pay out a significant percentage of their
earnings in the form of dividends and repurchases.
(2) Historically, dividends have been the predominant form of payout Share repurchases were relatively unimportant until the mid-1980s, but since then have
become an important form of payment.
(3) Among firms traded on organized exchanges in the USA, the proportion of
dividend-paying firms has been steadily declining Since the beginning of the
1980 s, most firms have initiated their cash payment to shareholders in the form
of repurchases rather than dividends.
(4) Individuals in high tax brackets receive large amounts in cash dividends and pay
substantial amounts of taxes on these dividends.
J Data on dividends and repurchases are from CRSP and Compustat Data on cash M&A activity (for
USA firms as acquirers only) is from SDC.
Ch 7:
Payout Policy
341
( 5) Corporations smooth dividends relative to earnings Repurchases are more volatile
than dividends.
( 6) The market reacts positively to announcements of repurchase and dividend
increases, and negatively to announcements of dividend decreases.
The challenge to financial economists has been to develop a payout policy
framework where firms maximize shareholders' wealth and investors maximize utility.
In such a framework payout policy would function in a way that is consistent with
these observations and is not rejected by empirical tests.
The seminal contribution to research on dividend policy is that of Miller and
Modigliani (1961) Prior to their paper, most economists believed that the more
dividends a firm paid, the more valuable the firm would be This view was derived
from an extension of the discounted dividends approach to firm valuation, which says
that the value V of the firm at date 0, if the first dividends are paid one period from
now at date 1, is given by the formula:
VO =
( + r)
(1)
where Dt = the dividends paid by the firm at the end of period t, and rt = the investors'
opportunity cost of capital for period t.
Gordon (1959) argued that investors' required rate of return rt would increase with
retention of earnings and increased investment Although the future dividend stream
would presumably be larger as a result of the increase in investment (i e , D, would
grow faster), Gordon felt that higher rt would overshadow this effect The reason for the
increase in r, would be the greater uncertainty associated with the increased investment
relative to the safety of the dividend.
Miller and Modigliani (1961) pointed out that this view of dividend policy is
incomplete and they developed a rigorous framework for analyzing payout policy They
show that what really counts is the firm's investment policy As long as investment
policy doesn't change, altering the mix of retained earnings and payout will not affect
firm's value The Miller and Modigliani framework has formed the foundation of
subsequent work on dividends and payout policy in general It is important to note
that their framework is rich enough to encompass both dividends and repurchases, as
the only determinant of a firm's value is its investment policy.
The payout literature that followed the Miller and Modigliani article attempted to
reconcile the indisputable logic of their dividend irrelevance theorem with the notion
that both managers and markets care about payouts, and dividends in particular The
theoretical work on this issue suggests five possible imperfections that management
should consider when it determines dividend policy:
(i) Taxes If dividends are taxed more heavily than capital gains, and investors cannot
use dynamic trading strategies to avoid this higher taxation, then minimizing
dividends is optimal.
342
E Allen and R Michaely
(ii) Asymmetric information If managers know more about the true worth of their
firm, dividends can be used to convey that information to the market, despite
the costs associated with paying those dividends (However, we note that with
asymmetric information, dividends can also be viewed as bad news Firms that
pay dividends are the ones that have no positive NPV projects in which to invest).
(iii) Incomplete contracts If contracts are incomplete or are not fully enforceable,
equityholders may, under some circumstances, use dividends to discipline
managers or to expropriate wealth from debtholders.
(iv) Institutional constraints If various institutions avoid investing in non or lowdividend-paying stocks because of legal restrictions, management may find that
it is optimal to pay dividends despite the tax burden it imposes on individual
investors.
(v) Transaction costs If dividend payments minimize transaction costs to equityholders (either direct transaction costs or the effort of self control), then positive
dividend payout may be optimal.
In Section 2 we elaborate further on some of the empirical observations about
corporate payout policies Section 3 reviews the Miller and Modigliani analysis.
Subsequent sections recount the literature that has relaxed their assumptions in various
ways.
2 Some empirical observations on payout policies
In the previous section we state six important empirical findings about corporate payout
policies Table 1 and Figure 1 illustrate the first observation that corporations pay out
a substantial portion of their earnings Table 1 shows that for USA industrial firms,
dollar expenditures on both dividends and repurchases have increased over the years.
The table also illustrates the second empirical observation above It shows that
dividends have been the dominant form of payout in the early period, but that
repurchases have become more and more important through the years For example,
during the 1970 S the average dividend payout was 38 % and the average repurchase
payout was 3 % By the 1990S the average dividend payout was 58 % and the
average repurchase payout was 27% From these numbers it appears that USA
corporations paid out over 80% of their earnings to shareholders 2 Clearly, payments
to shareholders through dividends and repurchases represent a significant portion of
corporate earnings However, we note that these numbers are tilted towards large firms
since we calculate payout as: ( Div/ E Earnings) In addition, aggregate earnings
(i.e , the denominator) contain many negative earnings This is especially true in the
later period, when more and more small, not yet profitable, firms registered on Nasdaq.
When we calculate payout for each firm and then average across firms (equal weighted)
2
See also Dunsby (1993) and Allen and Michaely (1995).
ICh
7:
Payout Policy
Ci
'0
W
vf
343
~
Ci
W~
m
N,
>
c
1::
C5
o\
~~NOP
~o~~ ~
H
N
m
v
\O
\~
\l
In
C
_ m
~
i
~
to,
C)
)
\
O\\
CsC
f)
O
O~f)
O
x
N
o\
em
o\ o
~
'0R
o\
m~
oo£Xmos
O
\
mm
o
e
o
o
CS
en
CA
o\
o\
A
N O£
WI
i
N
C)
O
C)
vO
q
'0
C
o
X
r
C
m
O
Ob
N
0
OqO\~
o
0
Cl
O\
nt
\~
_
O\
\
N
0
N
C
m
A
o~ a
0
.<
S
CC
N
N
00 '
o-m
~°
O
N
'IT
CC
00\o
Cn
\o\o CtD
m
~
I",
m
'IT
'0
~
'0
a,
ON
O
X
_ CO _'
N '
m
en
C> CIA In
O
N
O
~
In
O
In
en
'
a,
CS
"N O
'0 O
00
O
00
7
C
OO
00
O
t
'IT
O
t
°
\\
'00
X
e
'0o
v o\o
\°\
' O X
C
C 14 Cl
CA
_
l
oqo
'0\o
-2
00
°\
O
i
N
C
Pj
Cl
_
O
O
O
W) DIC
r
~
N
In
t
O
)
en
'0
O
N
00
O
m
N
rO
N
In
It
00
'
IT
'0
m)
1
N
In
N
'
00
It
CO
t
"'
'0
CtC
Ol
'0
'0
_t
O
'0
c
In
00
O
N
_)
_
NA '0,
'
C
O
m
'0
C
00
O
'
CC
l
O
O
O
H
C
D
00
to
a
CO
9-
O
N
'0
00
_
Cl
Cl
r
Cl
m
XMO
'
X
N
_
00
O
'0
O
'0
a,
OO
'0
Ct)
Cl
'0
N
Cl
C
O
O
~ ~ ~~D
ON
'
O
O
cl
O
'0
~
00
O
'0
t CN
t
In
C:t)
N
00
N
'0
O
'0
O
Cl
00
C
O
tO
)
O
OO
f
Cl
C
Cl
00
m
C,
'0
'0
00
N
00
m
Cl
00
t
'0
X
Ct)
Ct
XO
O
N
O
'O
N
NV 00
N
O
'
IN
-
V) o XcO___
Nn
X
C)
'
O
O
H
Cz
C
_
ng
I
CS
00
t
C)
)
m
CA
NC
Cl
N
O
Cl
OO
N
xO
O
O
=t
a,
C
o 00 '0N
o
00
O
CC
W,
CC
N
In
O
O
N
en
O
'0
cr
'0
0
00
N
O
O
a, O
n
Cl
N
0
r
C=
°
CC
°
_O
N
C
Cl
00
)
oo
x
C)
Cl
CO
C
O
CC
00
O
'm
'D
00
C
'Z
N
08
O
C Oa
CX
OO
r
'000
CC
c
CCC>
-t
'
Ct)
Om
N
'0
00
I
a,
t
O
00
'0
O
0CD
N
'0
~)
X
'0
O
'0E v
C
Cl
C
O
Cx
o
cs
'0O
eb
'O
OO
Cl
a
m
CCC
N
O
'0
'
00-m
O
O N
_
CCC CO
O
O,
'0
O
N
t
Cl
O
N
N
N
CN
O
'0
00
00
'
O
C
'0
m
C
N
00
CCC
'D
'
Ct
'
_
-r O
ots
-0
00
O
W
O
_
't
'0
Cl
00
C)
O ON
'0N
O
CC
C t
_os
)
'O
oO
Ct
°
'0
O
'0D
O
N
OO
N
O
O
l
Cl
'
'
C
N
O
'0
00
Cl N
_
00
t
O
CC
'CC
Ct)
Cl
00
O
C)
mt
' Oi
en
No
O
CC
D
O
CC
Clo
O
O O
D
O
om
O
'0
'
'
Cl
N
00
Ct
00
00
N
N
CC00 'm
O
Ct)
'0
N
Cl
l
cS
C
N
'O
N
XX- O
'0
Cl
o
o
'0
C
C
N>
_
N
_
N
In
'
N
_O
N
O
N
N
_
00
N
_
O
o
N
_
O
00
00
_O
In
00
_
00
_
O
00
_
O
_
'
00
aN
_O
00
00
_
00
00
_
O
CO
_
_
C
COE
C
0,
_O
E Allen and R Michaely
344
\
o
C \
\
o
,
1-
\
_
x
o
m
m
0)
O\
O\
00
O\
0
o
0
0
o
-
Pes
_
o
v*
e
O
oj
_
O
I
ce
i
i
00
0
0:
C:
CO
F-
omoo
0
o
\
O
x
b
m 00 O 00
m
__l
Nm O
*t)
~
O
N
I
'0
t'0
O
00
00-N
oo
O e _
N
N
N
\
N
00
0i
C
_m
N
'
O
o1
'r
N
.E
r
a,
W)
C
Cs
e
O
It
m
c
-0
-T
00
_
CC) 'N
C
l
l
°
WI)
*O
00
m~
oo
o
')
54.)
In
O
-
b
O
v
a,
In
Cl
0
Cl
'
-
00
m
r
0)
N
00
a,
O
O°
.
t
0
on
x
WC) W)
O
C
m
CC
"T
00
N
Cl
_____N
0)
N)
O
N
N
00
0,
0)
Ca,
a)
Ct
un
£c
r
C
0.)
0)
O
O
0)
.0,
Ch 7:
Payout Policy
345
Fig 1 Cash distributions to equityholders as a percentage of market value This figure depicts the average
total payout (dividends plus repurchases) yield, the average dividend yield, and the average repurchase
yield (all relative to market value) for a sample of USA firms The data sample consists of all firm-year
observations on Compustat (Full-Coverage, Primary, Secondary, Tertiary, Research, and Back Files) over
the period 1972-1998 that have positive earnings and have available information on the variables REPO,
DIV, and MV REPO is the expenditure on the purchase of common and preferred stocks (Compustat
item 115) minus any reduction in the value (redemption value) of the net number of preferred shares
outstanding (Compustat item 56) DIV is the total dollar amount of dividends declared on the common
stock (Compustat item 21) MV is the market value of common stock (Compustat item 24 x Compustat
item 25) The total payout is the sum of the dividend payout and the repurchase payout The data
sample contains 121973 firm-year observations and excludes banks, utilities, and insurance companies.
Based on data from Grullon and Michaely (2002), "Dividends, share repurchases and the substitution
hypothesis".
the overall payout relative to earnings is around 25 % lGrullon and Michaely ( 2002,
Figure 1)l.
To further illustrate the second observation, Figure 1 shows the evolution of dividend
yield (total dividends over market value of equity), repurchase yield (repurchases over
market value of equity) and payout yield (dividends plus repurchases over market value
of equity) since the early 1970 s Whether we examine repurchases relative to earnings
or to the market value of the firm, it is clear that repurchases as a payout method were
not a factor until the mid-1980 s It is interesting that in the 1990s, firms' average
total yield remained more or less constant while the dividend yield declined and the
repurchase yield increased.
The third observation is that dividends are now being paid by fewer firms As we
can see in Figure 2, Fama and French (2001) show that the proportion of firms that
pay dividends (among all CRSP-listed firms) has fallen dramatically over the years,
regardless of their earnings level Prior to the 1980 S firms that initiated a cash payment
346
E Allen and R Michaely
Fig 2 Percent of all CRSP firms in different dividend groups lFama and French (2001, Figure 2),
"Disappearing dividends: changing firm characteristics or lower propensity to pay?"l
usually did so with dividends But since the beginning of the 1980 s, most firms have
initiated cash payments with repurchases Figure 3 documents this observation for USA
industrial firms We define a cash distribution initiation as the first time after 1972 that
a firm pays dividends and/or repurchases shares Figure 3 shows that the proportion of
firms that initiated a cash distribution by using only share repurchases increased from
less than 27 % in 1974 to more than 81% in 1998 Share repurchase programs have
now become the preferred method of payout among firms initiating cash distributions
to their equityholders For earlier evidence on trend in repurchases see Bagwell and
Shoven (1989).
The fourth observation is that individuals pay substantial taxes on the large amounts
of dividends that they receive We collected information from the Federal Reserve's
Flow of Funds Accounts for the United States, and from the IRS, SOI Bulletin
about total dividends paid and the amounts received by individuals and corporations
for the years 1973-1996 Table 2 presents the results In most of the years in our
sample ( 1973-1996) individuals received more than 50% of the dividends paid out by
corporations Moreover, most of these dividends were received by individuals in high
tax brackets (those with annual gross income over $50000).
Peterson, Peterson and Ang (1985) conducted a study of the tax returns of
individuals in 1979 More than $33 b of dividends were included in individuals' gross
income that year The total of dividends paid out by corporations in 1979 was $57 7 b,
Ch 7:
347
Payout Policy
Cl.7 Pyu
0.8
4
olc
.
0.7
0.6 0.5
-
0.4
0.3
0.2
0.1 A
I I I I
1972 1975
I
Ir
I
I I
I I
I
I I
1978 1981 1984 1987 1990 1993 1996 1999
Year
Fig 3 Distribution of firms by payout method This figure depicts the distribution of firms by payout
method for a sample of USA firms We determine the payout policy of a firm by observing the cash
disbursements of the firm over a period of a year The data sample consists of all firm-year observations
on Compustat (Full-Coverage, Primary, Secondary, Tertiary, Research, and Back Files) over the period
1972-2000 that have available information on the following variables: REPO, DIV, EARN and MV REPO
is the expenditure on the purchase of common and preferred stocks (Compustat item 115) minus any
reduction in the value (redemption value) of the net number of preferred shares outstanding (Compustat
item 56) DIV is the total dollar amount of dividends declared on the common stock (Compustat
item 21) EARN is the earnings before extraordinary items (Compustat item 18) MV is the market
value of common stock (Compustat item 24 x Compustat item 25) The data sample contains 136646
firm-year observations and excludes banks, utilities, and insurance companies Squares, proportion of
firms that payout only with dividends; triangles, proportion of firms that payout with dividends and
repurchases; circles, proportion of firms that payout only with repurchases lGrullon and Michaely
(2002), "Dividends, share repurchases and the substitution hypothesis" l
so individuals received over two-thirds of that total The average marginal tax rate on
these dividends received by individuals (weighted by dividends received) was 40%.
The fact that individuals pay considerable taxes on dividends has been particularly
important in the dividend debate, because there appears to be a substantial tax
disadvantage to dividends compared to repurchases Dividends are taxed as ordinary
income Share repurchases are taxed on a capital gains basis Since the tax rate on
capital gains has usually been lower than the tax rate on ordinary income, investors
had an advantage if firms repurchased, rather than paid dividends Even after the 1986
Tax Reform Act (TRA) when the tax rates on ordinary income and capital gains were
equal for several years, there was a tax disadvantage to dividends because capital gains
were only taxed on realization In the 2001 tax code, long-term capital gains are lower
than ordinary income for most individual investors For example, an investor in the
348
F Allen and R Michaely
Table 2
Cash dividends from the corporate to the private sector
Year
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
(I)
a
0.774
0.740
0.727
0.741
0.718
0.696
0.708
0.710
0.690
0.653
0.624
0.600
0.572
0.592
0.578
0.617
0.612
0.617
0.630
0.620
0.611
0.585
0.579
0.543
0.513
0.485
0.495
(2 )b
29.9
33.2
33
39
44.8
50.8
57.7
64.1
73.8
76.2
83.6
91.0
97.7
106 3
112 2
129 6
155
165 6
178 5
185 5
203 2
234 9
254 2
297 7
333 7
348 6
364 7
(3)C
9.4
13.8
8.8
11.9
13.9
13.3
16.8
18.6
17.4
18.15
19.7
21.2
16.9
15.1
13.8
15.1
15.4
13.4
13.1
13.1
13.6
13.2
22.8
16.3
NA
NA
NA
(4 )d
18.7
20.8
21.9
24.5
27.8
30.2
33.5
43.6
48.1
52.1
48.6
48.6
55.0
61.6
66.8
77.3
81.3
80.2
77.3
77.9
79.7
(62 %)
(63 %)
(66%)
(63 %)
(62 %)
(59 %)
(58 %)
(68%)
(65 %)
(68%)
(58 %)
(53 %)
(56%)
(58 %)
(59 %)
(60 %)
(52 %)
(48 %)
(43 %)
(42 %)
(39%)
82.4 (35 %)
94.6 (37 %)
104 2 (35 %)
NA
NA
NA
e
(5)
42%
44%
45%
46%
47 %
50 %
53 %
54 %
52 %
55%
56 %
57 %
58 %
61 %
57 %
64%
66 %
66 %
66 %
67 %
65 %
66 %
71 %
73 %
NA
NA
NA
a Share of corporate equity owned by individuals Authors' calculation with data on market value of
domestic corporations and the holding (at market value) of households, personal trust and estates Source:
Table L 213 from the Federal Reserve statistical release, Flow of Funds Accounts of the United States,
March 2000.
b Total dividends paid by US corporations ($bln) From the Federal Reserve, Flow of Funds Accounts
of the United States, Table f 7, March 2000.
c Dividends received by corporations We include only dividends received from domestic corporations.
Internal Revenue Service, SOI Bulletin, Corporations return, Table 2, various years.
d Dividends received by individuals (% of total div) Internal Revenue Service, SOI Bulletin, Individuals
Tax Returns, Table 1 4, various years.
e Dividends received by individual with an adjusted gross income of over $50000 relative to dividends
received by all individual investors Source: Internal Revenue Service, SOI Bulletin, Individuals Tax
Returns, Table 1 4, various years.
Ch 7: Payout Policy
349
highest marginal tax bracket pays 39 6% taxes on dividends and only 20% tax on longterm capital gains Black ( 1976) calls the fact that corporations pay such large amounts
of dividends despite the existence of another, relatively untaxed, payout method, the
"dividend puzzle".
The fifth observation is that corporations smooth dividends From Table 1, we can
see that during the entire 1972-1998 period, aggregate dividends fell only twice (in
1992 and in 1998), and then only by very small amounts On the other hand, aggregate
earnings fell five times during the same time period and the drop was larger Unlike
dividends, repurchases are more volatile and more sensitive to economic conditions.
During the recession in the early 1970s, firms cut repurchases They did this again
during the recession of the early 1990 s Overall, between 1972 and 1998, aggregate
repurchases fell seven times.
Firms usually increase dividends gradually and rarely cut them Table 3 shows the
number of dividend increases and decreases for over 13 000 publicly held issues, for
the years 1971 to 2001 (Moody's dividend records, 1999 and S&P's dividend book,
2001) In each year, the number of dividend cuts is much smaller than the number
of dividend increases For example, in 1999, there were 1763 dividend increases or
initiations, but only 121 cuts or omissions.
In a classic study, Lintner (1956) showed that dividend-smoothing behavior was
widespread He started with over 600 listed companies and selected 28 to survey
and interview Lintner did not select these companies as a statistically representative
sample, but chose them to encompass a wide range of different situations.
Lintner made a number of important observations concerning the dividend policies
of these firms The first is that firms are primarily concerned with the stability of
dividends Firms do not set dividends de novo each quarter Instead, they first consider
whether they need to make any changes from the existing rate Only when they have
decided a change is necessary do they consider how large it should be Managers
appear to believe strongly that the market puts a premium on firms with a stable
dividend policy.
Second, Lintner observed that earnings were the most important determinant of any
change in dividends Management needed to explain to shareholders the reasons for
its actions, and needed to base its explanations on simple and observable indicators.
The level of earnings was the most important of these Most companies appeared to
have a target payout ratio; if there was a sudden unexpected increase in earnings, firms
adjusted their dividends slowly Firms were very reluctant to cut dividends.
Based on interviews of the 28 firms' management teams, Lintner reported a median
target payout ratio of 50 % Despite the very small sample and the fact that the study
was conducted nearly half a century ago, the target payout ratio is not far from what
we present in Table 1 for all USA industrial firms over a much longer time period.
Lintner's third finding was that management set dividend policy first Other policies
were then adjusted, taking dividend policy as given For example, if investment
opportunities were abundant and the firm had insufficient internal funds, it would resort
to outside funds.
350
E Allen and R Michaely
Table 3
Comparative annual dividend changes 1971-1993 (based on data from approximately 13200 publicly
held issues) a
Type of dividend change
Increase
Decrease
Resume
Omit
794
155
106
215
1972
1301
96
124
111
1973
2292
55
154
95
1974
2529
100
162
225
1975
1713
215
116
297
1976
2672
78
133
153
1977
3090
92
135
168
1978
3354
65
127
144
1979
3054
70
85
115
1980
2483
127
82
122
1981
2513
136
82
226
1982
1805
322
97
319
1983
1807
68
57
109
1984
1562
71
32
138
1985
1497
95
46
198
1986
1587
71
54
107
1987
1702
65
40
117
1971
1988
1683
80
42
152
1989
1312
137
39
255
1990
1072
188
48
264
1991
1314
139
55
145
1992
1333
131
53
146
1993
1635
87
75
106
1994
1826
59
52
77
1995
1882
49
51
73
1996
2171
50
37
80
1997
2139
46
24
49
1998
2047
84
17
61
1999
1701
62
38
83
2000
1438
69
32
75
2001
1244
117
17
70
a For data until 1982, Moody's Dividend Record; for data between 1983 and 2001, S&P dividend
record.
Ch 7:
351
Payout Policy
Lintner suggested that the following model captured the most important elements
of firms' dividend policies For firm i,
Di*= aiEit,
Dit Di(t 1) = ai + ci(D*t
( 2)
Di(t 1)) +uit,
(3)
where, for firm i, D* = desired dividend payment during period t; Di, = actual dividend
payment during period t; a = target payout ratio; Ei = earnings of the firm during
period t; ai = a constant relating to dividend growth; ci = partial adjustment factor;
ui, = error term This model was able to explain 85 % of the dividend changes in his
sample of companies.
Fama and Babiak (1968) undertook a comprehensive study of the Lintner model's
performance, using data for 392 major industrial firms over the period 1946 through
1964 They also found the Lintner model performed well Over the years, other studies
have confirmed this.
The sixth observation is that the market usually reacts positively to announcements
of increases in payouts and negatively to announcements of dividend decreases This
phenomenon has been documented by many studies, such as Pettit (1972), Charest
(1978), Aharony and Swary (1980) and Michaely, Thaler and Womack (1995) for
dividends, and by Ikenberry, Lakonishok and Vermaelen (1995) for repurchases.
This evidence is consistent with managers knowing more than outside shareholders,
and dividends and repurchases changes provide some information on future cash
flows le g , Bhattacharya (1979), or Miller and Rock ( 1985), or about the cost of
capital lGrullon, Michaely and Swaminathan (2002), Grullon and Michaely (2003)l.
The evidence is also consistent with the notion that when contracts are incomplete,
higher payouts can sometimes be used to align management's interest with that of
shareholders', as suggested by Grossman and Hart ( 1980), Easterbrook ( 1984) and
Jensen (1986).
3 The Miller-Modigliani dividend irrelevance proposition
Miller and Modigliani ( 1961) showed that in perfect and complete capital markets, a
firm's dividend policy does not affect its value The basic premise of their argument
is that firm value is determined by choosing optimal investments The net payout is
the difference between earnings and investment, and is simply a residual Because
the net payout comprises dividends and share issues/repurchases, a firm can adjust
its dividends to any level with an offsetting change in shares outstanding From the
perspective of investors, dividend policy is irrelevant, because any desired stream
of payments can be replicated by appropriate purchases and sales of equity Thus,
investors will not pay a premium for any particular dividend policy.
352
E Allen and R Michaely
To illustrate the argument behind the theorem, suppose there are perfect and
complete capital markets (with no taxes) At date t, the value of the firm is
V, = present value of payouts,
where payouts include dividends and repurchases For ease of exposition, we initially
consider the case with two periods, t and t + 1 At date t, a firm has
earnings, Et, (earned previously) on hand.
It must decide on
the level of investment, I,
the level of dividends, Dt
the amount of shares to be issued, A St (or repurchased if A St is negative).
The level of earnings at t + 1, denoted Et+,(It,,t+,), depends on the level of
investment It, and a random variable t,+1 Since t + 1 is the final date, all earnings
are paid out at t + 1 Given complete markets, let
pt(ot+ ) = time t price of consumption in state ot+1.
Then it follows that
t = D, A St +
p(t+ ) E+ (It, t + 1) dt +
(4)
The sources and uses of funds identity says that in the current period t:
Et +A St = I, + D
(5)
Using this to substitute for current payouts, D, AS, gives
Vt
= Et -It +
f pt(t + )Et +
(I,
Ot,)
d
+1
(6)
From Equation (6) we can immediately see the first insight from Miller and
Modigliani's analysis Since Et is given, the only determinant of the value of the firm
is current investment It.
This analysis can be extended to the case with more than two periods Now
Vt = E
It + t + ,
(7)
where
Vt+ 1 = Et+ (lt, Ot+ 1) -I,+1 + Vt+ 2,
(8)
and so on, recursively It follows from this extension that it is only the sequence of
investments It, It+,,
that is important in determining firm value Firm value is
maximized by making an appropriate choice of investment policy.
Ch 7:
Payout Policy
353
The second insight from the Miller-Modigliani analysis concerns the firm's dividend
policy, which involves setting the value of D, each period Given that investment
is chosen to maximize firm value, the firm's payout in period t, D, AS,, must be
equal to the difference between earnings and investment, E, I, However, the level of
dividends, D,, can take any value, since the level of share issuance, AS,, can always
be set to offset this It follows that dividend policy does not affect firm value at all It
is only investment policy that matters.
The analysis above implicitly assumes 100 % equity financing It can be extended to
include debt financing In this case management can finance dividends by using both
debt and equity issues This added degree of freedom does not affect the result As with
equity-financed dividends, no additional value is created by debt-financed dividends,
since capital markets are perfect and complete so the amount of debt does not affect
the total value of the firm.
The third and perhaps most important insight of Miller and Modigliani's analysis
is that it identifies the situations in which dividend policy can affect firm value It
could matter, not because dividends are "safer" than capital gains, as was traditionally
argued, but because one of the assumptions underlying the result is violated.
Perfect and complete capital markets have the following elements:
(1) No taxes.
(2) Symmetric information.
(3) Complete contracting possibilities.
(4) No transaction costs.
(5) Complete markets.
It is easy to see the role played by each of the above assumptions The reason for
Assumption 1 is clear In the no-taxes case, it is irrelevant whether a firm pays out
dividends or repurchases shares; what is important is D, -AS, If dividends and share
repurchases are taxed differently, this is no longer the case Suppose, for example,
dividends are taxed at a higher rate than capital gains from share repurchases Then it is
optimal not to pay dividends, but instead to pay out any residual funds by repurchasing
shares In Section 5 we discuss the issues raised by relaxing Assumption 1.
Assumption 2 is that all participants (including the firms) have exactly the same
information set In practice, this is rarely the case Managers are insiders and are
likely to know more about the current and future prospects of the firm than outsiders.
Dividends can reveal some information to outsiders about the value of the corporation.
Moreover, insiders might even use dividends to deliberately change the market's
perception about the firm's value Again, dividend policy can affect firm value Sections
6.1 and 7 1 consider the effect of asymmetric information.
The complete contracting possibilities specified in Assumption 3 mean that there
is no agency problem between managers and security holders, for example In this
case, motivating the decisions of managers is possible through the use of enforceable
contracts Without complete contracting possibilities, dividend policy could, for
example, help ensure that managers act in the interest of shareholders A high payout
354
E Allen and R Michaely
ratio causes management to be more disciplined in the use of the firm's resources and
consequently increase firm value We cover these issues in Sections 6 2 and 7 2.
Assumption 4 concerns transaction costs These come in a variety of forms For
example, firms can distribute cash through dividends and raise capital through equity
issues If flotation costs are significant, then every trip to the capital market will reduce
the firm's value This means changing dividend policy can change the value of the firm.
By the same token, when investors sell securities and make decisions about such sales,
the transaction costs that investors incur can also result in dividend policy affecting
the value of the firm Section 8 develops several transaction-cost-related theories of
dividend policy.
Assumption 5 is that markets are complete To illustrate why this is important,
assume that because trading opportunities are limited, there are two groups with
different marginal rates of substitution between current and future consumption By
adjusting its dividend policy, a firm might be able to increase its value by appealing to
one of these groups The literature has paid very little attention to explanations such as
these for dividend policy Nevertheless, these explanations could be important if some
investors wish to buy stocks with a steady income stream, and markets are incomplete
because of high transaction costs Further analysis in this area might provide some
insights into dividend policy.
Another issue that is central to our survey is the form of the payout One area of
significant growth in the literature is related to the role of repurchases as a form of
payout, not only because repurchases have become more popular (Table 1), but also
because of the research concerning the reasons for repurchases and the interrelation
between dividends and repurchases In Section 4 we define corporate payout, both
conceptually and empirically In Section 9 we review in detail the recent developments
concerning repurchases.
4 How should we measure payout?
The Miller and Modigliani framework defines payout policy as the net payout to
shareholders However, most empirical work measures payout only by the amount of
dividends the firms pay Such studies do not consider repurchases Neither do they
factor in either net payout (accounting for capital raising activities) or cash spent on
mergers and acquisitions.
If we wish to find out how much cash corporations pay out (relative to their earnings)
at the aggregate level, we need to consider some of the aggregate measures, such as
the one presented in Table 1, namely, aggregate dividends plus aggregate repurchases
relative to aggregate earnings But even this measure is incomplete First, shareholders
also receive cash payouts from corporations through mergers and acquisitions that are
accomplished through cash transactions That is, shareholders of the acquired firms
receive a cash payment that can be viewed as a liquidating (or final) dividend.
Ch 7:
355
Payout Policy
Table 4
Mergers and acquisitions and capital raising activities by USA corporationsa
Year
(1) Total M&A
activity ($mln)
(2) Cash
mergers (where
USA firms
are the target)
(3) IPOs
proceeds ($mln)
(4) SE Os
proceeds ($mln)
(5) Net payout
from M&A and
raising capital
(2 3 -4)
221
382
-412
1978
8882
8086
225
305
7556
1979
7993
7589
398
247
6944
1977
191 8
191 8
1980
17570
10417
1387
10901
-1871
1981
86098
59725
3114
10958
45653
1982
53426
27080
1339
14743
10998
1983
82757
30539
12460
26071
-7992
1984
151709
94029
3868
6032
84129
1985
169156
151999
8477
16493
127029
1986
193620
167028
22251
20430
124347
1987
185730
158662
23982
16613
118067
1988
310895
289377
23806
5941
259630
1989
235759
194966
13706
9332
171928
1990
143402
109427
10122
8998
90307
1991
106659
66778
25138
33749
7890
1992
130264
75957
39620
31866
4471
1993
203545
113186
57423
48995
6768
1994
307047
183956
33728
27487
122741
1995
462829
228104
30207
54176
143721
1996
544484
306812
50000
71222
185590
1997
819663
390359
44226
75409
270724
1998
1392997
410619
43721
70886
296012
1999
1021026
543324
71327
100048
371949
a Thompson Financial Securities Data.
Using data from SDC, Table 4 presents the magnitude of such payments For each
year we calculate the total dollar amount that was paid to USA corporations in all
cash M&A deals (Note that this figure is a lower bound, since it does not account for
deals in which payment was partially in cash and partially in stocks) The amount is
not trivial and it does vary by year This type of liquidating dividend seems to have
a significant weight in the aggregate payout of USA corporations For example, in
1999, proceeds from cash M&As were more than the combined cash distributed to
shareholders through dividends and repurchases combined.
356
E Allen and R Michaely
Our next measure accounts not only for the outflow of funds from corporations
to their shareholders, but also for the inflow of funds Columns 3 and 4 in Table 4
present the dollar amount of capital raised by USA corporations through SE Os and
IP Os Column 5 reports the net amount (cash from M&As minus proceeds from IP Os
and SEOs) It is clear that these are significant amounts When we compare Tables
1 and 4, we see that in the last decade these amounts are as large as the cash payments
through dividends and repurchases combined We are also interested to see its impact
on the overall aggregate payout Clearly, in some years the aggregate payout is higher
than after-tax earnings.
One can also define the aggregate payout as the total transfer of cash from the
corporate sector to the private sector This definition contains three elements: dividends
paid to individual investors, repurchase of shares from individual investors, and net
cash M&A activity where the proceeds are going to the private sector.
Using this definition and information from the IRS Statistics of Income and the
Federal Reserve Flow of Funds publications, we can recalculate a rough measure
of the total payout to the private sector over the years We base this measure on the
total dividends, repurchases, and cash M&A activity We assume that the proportional
holdings of each group (individuals, corporations and institutions) are the same for all
firms in the economy.
In Table 2, we calculate the portion of shares held by individual investors (using
information from Table L-312 from the Federal Reserve Flow of Funds) 3 Using
this ratio, we can approximate the portion of repurchased shares and net cash M&As
that went to the private sector For example, in 1995, the private sector received
$94b in dividends (see Table 2), $82 b in cash M&As (57 9 % of shares owned by
individuals multiplied by $143 b of net cash M&As, see Tables 2 and 4), and roughly
$50b in repurchases (57 9 % of shares owned by individuals multiplied by $72 3 b of
repurchases; see Tables 1 and 2) We note that out of total cash payments to the private
sector of around $219b, less than half is through "formal" dividends Table 5 presents
the cash payout that goes to the private sector (dividends, repurchases, and net cash
M&As) for the various years.
These issues have not received much attention in the literature We believe they
should It is difficult to take a position on payout policy before we correctly measure
it.
An equally interesting issue is to analyze the payout, its components, and the
relation between payout and earnings at the firm level For example, we think it would
be interesting to investigate the type of firm that gives its shareholders liquidating
dividends, and how such dividends relate to other types of payout Analyzing the
interaction between total payout, dividends, and the recent surge in repurchases would
3 Total dividends are taken from Table F-7 (distribution of national income) of the Flow of Funds
Accounts of the USA The portion of dividends received by individuals is from Table 1 of the
SOI Bulletin, Winter 1999-2000.
Ch 7:
357
Payout Policy
Table 5
Net total payout to individual investors
(8) g
Year
(2) a
(3) b
(4) c
(5) d
(6) e
(7 )f
1977
0 718
-412
-296
3566
2560
27800
30065
1978
0 696
7556
5259
4256
2962
30200
38421
1979
0 708
6944
4916
5421
3838
33500
42254
1980
0 71
-1871
-1328
5689
4039
43600
46311
1981
0 69
45653
31501
6262
4321
48100
83921
1982
0 653
10998
7182
9593
6264
52100
65546
1983
0 624
-7992
-4987
8899
5553
48600
49166
1984
06
84129
50477
27971
16783
48600
115860
1985
0 572
127029
72661
33136
18954
55000
146614
1986
0 592
124347
73613
35707
21139
61600
156352
1987
0 578
118067
68243
52041
30080
66800
165122
1988
0 617
259630
160192
48765
30088
77300
267580
1989
0 612
171928
105220
54949
33629
81300
220149
1990
0 617
90307
55719
46275
28552
80200
164471
1991
0 63
7890
4971
22962
14466
77300
96737
1992
0 62
4471
2772
33289
20639
77900
101311
1993
0 611
6768
4135
36334
22200
79700
106035
1994
0 585
122741
71803
46503
27204
82400
181408
1995
0 579
143721
83214
72330
41879
94600
219694
1996
0 543
185590
100775
101808
55282
104200
260257
1997
0 513
270724
138881
143842
73791
NA
NA
1998
0 485
296012
143566
175488
85112
NA
NA
1999
0 495
371949
184115
202000
99990
NA
NA
a Portion held by individuals (from Table 2).
b Net payout from M&A and raising capital (from Table 4).
c Net M&A payout to individual investors (column 2xcolumn 3).
d Amount repurchased (Table 1).
e Amount repurchased from individual investors (column 2 x column 5).
f Dividends received by individuals (from Table 2).
g Net total payout to individual investors (columns 4 + 6 + 7).
also require information on individual firms' payout policies But at the firm level,
there may be another problem in the definition of payout relative to earnings, since a
significant portion of firms have negative earnings For these firms, it is not possible
to define a total payout ratio, a repurchase payout ratio, or a dividend payout ratio.
Our discussion highlights several important points First, in our opinion, the main
issue is not whether one measure is better than another Instead, we ask, what is the
358
E Allen and R Michaely
question that we are trying to answer? This question in turn should have an impact on
which definition of payout we use.
The issue of how to define payout is also very relevant to the excess volatility
literature For example, Ackert and Smith (1993) showed that the results of variancebound tests depend on how we measure cash distributions to shareholders When they
used only stated dividends, they found evidence of excess volatility When the payout
measure included share repurchase and takeover distributions as well, they did not find
evidence of excess volatility It is likely that using the net total payout to investors will
add some variability to cash flows It may also reduce even further the discrepancy
between cash flow volatility and price volatility In our opinion, this issue is worthy
of further research.
Second, it is clear that most of the finance literature has analyzed the payout policy
question using only the very narrow definition of dividend payout Some studies have
attempted to analyze repurchase payout But with only a few exceptions, the literature
does not cover the issue of total payout, its composition, and determination This
lacuna is understandable, given the fact that over many years, dividends were the most
prominent form of payout But this is not so anymore Thus, to a great extent our review
article reflects the current literature We devote more space and put more emphasis on
dividends relative to the other forms of payouts We hope future research will explore
the other aspects of payout policy and their implications.
5 Taxes
Much of the literature on payout policy focuses on the importance of taxes, and tries
to reconcile several of the empirical observations discussed in our introduction Firms
pay out a large part of their earnings as dividends; many of the recipients are in high
tax brackets Firms did not traditionally use repurchases as a method of payout The
basic aim of the tax-related literature on dividends has been to investigate whether
there is a tax effect: All else equal, we ask if firms that pay out high dividends are
less valuable than firms that pay out low dividends.
Two basic ideas are important to understanding how to interpret the results of these
investigations:
( 1) Static clientele models:
(i) Different groups, or "clienteles", are taxed differently Miller and Modigliani
(1961) argued that firms have an incentive to supply stocks that minimize the
taxes of each clientele In equilibrium, no further possibilities for reducing
taxes will exist and all firms will be equally priced.
(ii) A particular case (labeled as the simple static model) is when all investors are
taxed the same way, and capital gains are taxed less than dividend income In
this case, the optimal policy is not to pay dividends Firms with high dividend
yields would be worth less than equivalent firms with low dividend yields.
Ch 7:
Payout Policy
359
(2) Dynamic clientele model: If investors can trade through time, tax liabilities can
be reduced even further The dividend-paying stock will end up (just before the
ex-dividend day) in the hands of those who are taxed the least when the dividend
is received Such trades will be reversed directly after the ex-day.
The empirical studies of dividend policy have tried to distinguish between the
different versions of these models by attempting to identify one or more of the
following:
(i) Is there a tax effect so that low-dividend-paying stocks are more valuable than
high dividend stocks?
(ii) Do static tax clienteles exist so that the marginal tax rates of high-dividend
stockholders are lower than those of low-dividend stockholders?
(iii) Do dynamic tax clienteles exist so that there is a large volume around the exdividend day, and low-tax-rate investors actually receive the dividend?
This literature has traditionally been divided into CAPM-based studies and exdividend day studies In our view, more insight is gained by comparing static to
dynamic models In the static models, investors trade only once Thus, with the
objective of minimizing taxes (keeping all else constant), investors must make a
long-term decision about their holdings The buy-and-hold CAPM studies, such as
Litzenberger and Ramaswamy ( 1979) and Miller and Scholes (1982), fall into this
category The Elton and Gruber (1970) study is similar in that respect Investors are
allowed to trade only once, either on the cum-day or on the ex-day, but not on both.
As we shall show, a static view is appropriate when transaction costs are exceedingly
high, or when tax payments have been reduced to zero in the static clientele model.
In contrast, in dynamic models, investors are allowed to take different positions at
different times These models take into account risk, taxes, and transaction costs Just
before the ex-day, dividend-paying stocks can flow temporarily to the investors who
value them the most.
5.1 Static models
First, we look at the special case in which all investors are taxed in the same way
and the tax rate on dividend income is higher than the tax rate on capital gains
income In otherwise perfect capital markets, the optimal policy is to pay no dividends.
Equityholders are better off receiving profits through repurchases or selling their shares
so that they pay capital gains taxes rather than the higher taxes on dividends Most USA
corporations have not followed this scenario For a long time, many firms have paid
dividends regularly and have rarely repurchased their shares On the face of it, this
behavior is puzzling, especially if we believe that agents in the market place behave
in a rational manner The basic assumption of this simple static model is that for all
investors there is a substantial tax disadvantage to dividends because they are taxed
(heavily) as ordinary income, while share repurchases are taxed (lightly) as capital
gains.
360
E Allen and R Michaely
But even if the statutory tax rates on dividends and capital gains were equal (and
usually, they have not been), from a tax perspective receiving unrealized capital gains
is superior to dividend payments.
The first reason is that capital gains do not have to be realized immediately, and thus
the associated tax can be postponed An investor's ability to postpone may generate
considerable value Imagine a stock with an expected annual return of 15%, and an
investor with a marginal tax rate of 20 % on long-term capital gains Say the investor
has $1000 and an investment horizon of ten years, and consider whether she should
realize gains at the end of each year or wait and realize all gains at the end of the
tenth year Under the first strategy, her final wealth would be $3106 Under the second
strategy it would be $3436, a substantial difference.
Second, investors can choose when to realize capital gains (unlike dividends, for
which they have no choice in the timing) In a more formal setting Constantinides
( 1984) showed that investors should be willing to pay for this option to delay capital
gains realization, and labeled it the "tax timing option".
In reality, of course, not all investors are taxed as individuals Many financial
institutions, such as pension funds and endowments, do not pay taxes They have
no reason to prefer capital gains to dividends, or vice versa Individuals hold stocks
directly or indirectly, and so do corporations One of the principal reasons corporations
hold dividend-paying stocks as both a form of near-cash assets and as an investment
is because under the USA tax code, a large fraction of intercorporate dividends are
exempt from taxation, but intercorporate (or government) interest payments are not.
Under the old tax code, only 15 % of dividends, deemed taxable income, were taxed,
so the effective tax rate on dividends received was 0 15 x 0 46 (marginal corporate tax
rate) = 6 9 % But corporations had to pay the full amount of taxes on any realized
gains Under the current tax code, 30% of dividends are taxed 4
In a clientele model, taxpayers in different groups hold different types of assets,
as illustrated in the stylized example below Individuals hold low-dividend-payout
stocks Medium-dividend-payout firms are owned by people who can avoid taxes, or
by tax-free institutions Corporations own high-dividend-payout stocks Firms must be
indifferent between the three types of stock, or they would increase their value by
issuing more of the type that they prefer.
4 Prior to the 1986 Tax Reform Act (TRA), individual investors who held a stock for at least six
months paid a lower tax on capital gains (20%) than on ordinary dividends (50 %) The TRA eliminated
all distinctions between capital gains and ordinary income However, it is still possible to defer taxes
on capital gains by not realizing the gains Before the 1986 TRA, a corporation that held the stock
of another corporation paid taxes on only 15% of the dividend Therefore, the effective tax rate for
dividend income was 0 15 x O 46 = 0 069 After the TRA, the corporation income tax rate was reduced
to 34 % The fraction of the dividend exempted from taxes was also reduced to 70 % The effective tax
rate for dividend income was therefore increased to 0 3 x 0 34 = O102 In both time periods, the dividend
exemption could be as high as 100% if the dividend-paying corporation was a wholly owned subsidiary
of the dividend-receiving corporation.
Ch 7:
361
Payout Policy
Table 6
A clientele model example
Dividend payout
Before-tax earnings/share
High
Medium
Low
$100
$100
$100
$100
$50
$0
$0
$50
$100
Payout policy:
Dividends
Capital gains
After-tax payoff/share for group:
(i) Individuals
$50
$65
$80
(ii) Corporations
$90
$77 5
$65
(iii) Institutions
Equilibrium price/share
$100
$100
$100
$1000
$1000
$1000
How are assets priced in this model? Since firms must be indifferent between the
different types of assets, the assets must be priced so they are equally desirable To
show how this works, we use the following example.
Suppose there are three groups that hold stocks:
(i) Individuals who are in high tax brackets and pay high taxes on dividend-paying
stocks These investors are subject to a 50% tax rate on dividend income and a
20 % tax rate on capital gains.
(ii) Corporations whose tax situation is such that they pay low taxes on stocks that
pay dividends Their tax rate on dividend income is 10 % and 35 % on capital
gains.
(iii) Institutions that pay no taxes Their opportunity cost of capital, determined by
the return available in investment other than securities, is 10%.
Assume that these groups are risk neutral, so risk is not an issue All that matters is
the after-tax returns to the stocks (We note that in this stylized market, a tax clientele
is a result of both the risk neutrality assumption and the trading restrictions).
There are three types of stock For simplicity, we assume that each stock has earnings
per share of $100 The only difference between these shares is the form of payout.
Table 6 describes the after-tax cash flow for each group if they held each type of
stock.
In this example, individuals with high tax brackets will hold low-payout shares,
corporations will hold the high-payout shares, and institutions will be prepared to hold
all three The asset holdings of these three groups are shown in Table 7.
To show why the shares must all have the same price, if the price of low-payout
shares was $1050 and the prices of the high and medium-payout stocks was $1000,
what would happen? High and medium-payout firms would have an incentive to
change their dividend policies and increase the supply of low-payout stocks This
F Allen and R Michaely
362
Table 7
Asset holdings in the clientele model example
Group
Asset holdings
High tax bracket
Low-dividend-payout assets
Corporations
High-dividend-payout assets
Tax-free institutions
Any assets
change would put downward pressure on the price of low-payout stock What amount
of stock do investors demand? Individuals would still be prepared to buy the lowpayout stock, since $80/$1050= 7 62%, which is greater than the 6 5 % ($65/$ 1000)
they would obtain from holding medium-payout stocks, or the 5 % ($50/$ 1000) they
would obtain from holding high-payout stocks What about institutions? They will not
be prepared to hold low-payout stocks, since the return on them is $ 100/$ 1050 = 9 52%.
This return is less than the 10 % ($ 100/$ 1000) they can get on the other two stocks and
the opportunity cost they obtain from holding foreign assets, so they will try to sell.
Again, there is downward pressure on the price of low-payout stock Therefore, the
price must fall from $1050 to $1000 for equilibrium to be restored A similar argument
explains why the prices of other stocks are also $1000 Thus, in equilibrium, the price
is independent of payout policy and dividend policy is irrelevant, as in the original
Miller and Modigliani theory 5
Several studies have attempted to distinguish between the case of the static model in
which everybody is taxed the same, and the static clientele model in which investors
are taxed differently Perhaps the easiest way to make the distinction is to investigate
the relation between the marginal tax rates of stockholders and the amount of dividends
paid.
Blume, Crockett and Friend (1974) found some evidence from survey data that there
is a modest (inverse) relation between investors' tax brackets and the dividend yield of
the stocks they hold Lewellen, Stanley, Lease and Schlarbaum (1978), using individual
investor data supplied by a brokerage firm, found very little evidence of this type of
effect Both studies indicate that investors in high tax brackets hold substantial amounts
of dividend-paying stock.
Table 2 corroborates these findings for the last 30 years It is evident that individuals
in high tax brackets hold substantial amounts of dividend-paying stocks There is no
evidence that their dividend income relative to capital gains income is lower than that
of investors in low tax brackets According to the clientele theory, this phenomenon
5 The equilibrium here is conceptually different from the one in Miller (1977) Miller presents an
equilibrium in which there is a strict clientele In the equilibrium here, potential arbitrage by institutions
ensures one price for all stocks, regardless of their dividend policy The existence of a strict tax-clientele
is inconsistent with no-arbitrage See also Blume (1980).
Ch 7:
363
Payout Policy
should not occur For example, firms should be able to increase their value by switching
from a policy of paying dividends to repurchasing shares.
Elton and Gruber ( 1970) sought to identify the relation between marginal tax rates
and dividend yield by using ex-dividend date price data They argued that when
investors were about to sell a stock around its ex-dividend date, they would calculate
whether they were better off selling just before it goes ex-dividend, or just after If
they sold before the stock went ex-dividend, they got a higher price Their marginal
tax liability was on the capital gain, represented by the difference between the two
prices If they sold just after, the price would have fallen because the dividend had
been paid They would receive the dividend plus this low price, and their marginal tax
liability would be their personal tax rate times the dividend In this setting, we can make
a direct comparison between the market valuation of after-tax dividend dollars and
after-tax capital gains dollars In equilibrium, stocks must be priced so that individuals'
marginal tax liabilities are the same for both strategies.
Assuming investors are risk neutral and there are no transaction costs, it is necessary
that:
PB tg(PB -PO) = PA
tg(PA
P)
+ D( 1 td),
(9)
where PB = stock price cum-dividend (the last day the stock is traded with the
dividend); PA = expected stock price on the ex-dividend day (the first day the stock
is traded without the dividend); Po = stock price at initial purchase; D = dividend
amount; tg = personal tax rate on capital gains; td = personal tax rate on dividends.
The left-hand side of Equation (9) represents the after-tax receipts the seller would
receive if he sold the stock cum-dividend and had bought it originally for PO The
right-hand side represents the expected net receipts from sale on the ex-dividend day.
Rearranging,
PB PA
D
1 td
( 0)
l-tg
If there are clienteles with different tax brackets, the tax rates implied by the ratio of
the price change to the dividend will differ for stocks with different levels of dividends.
The implied tax rate will be greater the higher the dividend yield, and, hence, the lower
the tax bracket of investors Elton and Gruber find strong evidence of a clientele effect
that is consistent with this relation.
5.1 1 The role of risk
In the simplest versions of the theories presented above, risk has been ignored In
practice, because risk is likely to be of primary importance, it must be explicitly
incorporated in the analysis.
As Long (1977) pointed out, there is an implicit assumption in the argument of
a tax clientele that when there is risk, there are redundant securities in the market.
364
E Allen and R Michaely
An investor can achieve the desired portfolio allocation in risk characteristics without
regard to dividend yield In other words, investors can create several identical portfolios
in all aspects but dividend yield.
Keim (1985) presented evidence that stocks with different yields also have different
risk characteristics Zero-dividend-yield stocks and stocks with low dividend yields
have significantly higher betas than do high-yield stocks This finding implies that
it may be a nontrivial task to choose the optimal risk-return tradeoff while ignoring
dividend yield.
Depending on the precise assumptions made, some models that incorporate risk are
similar to the simple static model, in that there is a tax effect and dividend policy
affects value On the other hand, other models are similar to the static clientele model
in that there is no tax effect and dividend policy does not affect value Therefore, most
of the literature has focused on the issue of whether or not there is a tax effect.
Brennan ( 1970) was the first to develop an after-tax version of the CAPM.
Litzenberger and Ramaswamy (1979, 1980) extend his model to incorporate borrowing
and short-selling constraints In both cases, the basic result is that for a given level of
risk, the compensation for a higher dividend yield is positively related to the differential
taxes between dividends and capital gains:
E(Rit Rfi) = al +a2 At +a3(dit -Rfi)
(11)
Equation (11) describes the equilibrium relation between a security's expected return
E(Ri,), its expected dividend yield (di,), and its systematic risk (fit) Finding a
significantly positive a 3 is interpreted as evidence of a tax effect That is, two stocks
with the same risk exposure (same beta) will have the same expected return only if
they have the same dividend yield Otherwise, the stock with the higher dividend yield
will have a higher expected return to compensate for the higher tax burden associated
with the dividend.
Several researchers have tested such a relation, including Black and Scholes (1974),
Blume ( 1980), Morgan ( 1982), Poterba and Summers ( 1984), Keim (1985), Rosenberg
and Marathe ( 1979), Miller and Scholes (1982), Chen, Grundy and Stambaugh (1990)
and Kalay and Michaely (2000) The empirical results are mixed Several of these
studies find a positive yield coefficient, which they attribute to differential taxes.
Black and Scholes (1974) performed one of the earliest (and one of the most
influential) tests Using annual data, and a slightly different version of Equation (11),
they tested the tax effect hypothesis:
Ri = Yo + lRm
ol
i + Yl (di dm)/dm +i,
i = 1,
N,
(12)
where Pl = the rate of return on the ith portfolio; yo = an intercept term that should
be equal to the risk-free rate, Rf, based on the CAPM; Rm = the rate of return on the
market portfolio; fi = the systematic risk of the ith portfolio; Yl = the dividend impact
coefficient; di = the dividend yield on the ith portfolio, which is measured as the sum
365
Ch 7: Payout Policy
of dividends paid during the previous year divided by the end-of-year stock price; dm =
the dividend yield on the market portfolio measured over the prior 12 months; ei =
the error term.
To test the tax effect, Black and Scholes formed portfolios of stocks and used a longrun estimate of dividend yield (the sum of prior-year dividends divided by year-end
price) Their null hypothesis was that the dividend-yield coefficient is not significantly
different from zero This hypothesis cannot be rejected for the entire time period (1936
through 1966) or for any of the ten-year subperiods Black and Scholes concluded that
"
it is not possible to demonstrate that the expected returns on high yield common
stocks differ from the expected return on low yield common stocks either before or
after taxes".
In a series of studies, Litzenberger and Ramaswamy (1979, 1980, 1982) re-examined
this issue 6 Their experimental design differs from that of Black and Scholes (1974)
in several important aspects They use individual instead of grouped data, and they
correct for the error in variables problems in the beta estimation by using maximum
likelihood procedures Perhaps most important, they classify stock into yield classes
by using a monthly definition of dividend yield, rather than a long-term dividend yield
definition as in Black and Scholes (1974).
The Litzenberger and Ramaswamy experiment involves three steps First, they
estimate the systematic risk of each stock for each one of the test months The
estimation uses the market model regression Formally,
Ri, -RRt = ai, +Pit(Rmj -Rfj)+it j = t-60,
1,
,t
( 13)
where Rmj is the return on the market portfolio during periodj; R is the rate of return
on stock i during period j; fli, is the estimated beta for stock i for period t; the riskless
rate of interest during period t is Rft; and eit is a noise term The second stage uses
the estimated beta for stock i during month t, pit, and an estimate of stock i's expected
dividend yield for month t, di, as independent variables in the following cross-sectional
regression for month t:
Rit
Rf = alt +a 2tlit +a3t(dit Rf) + i i
=
N
(14)
The experiment requires an ex-ante estimate of the test month dividend yield They
obtain the estimate of expected dividend yield for month t from past observations For
cases in which the dividends are announced at month t 1, the estimate is d/pt_ 1.
When the announcement and ex-date occur in month t, Litzenberger and Ramaswamy had to estimate the market's time t expected dividend as of the end of
month t 1 The estimate they chose was the last dividend paid during the previous
6 The econometric technique used by Litzenberger and Ramaswamy to correct for the errors in variables
problem represents a significant contribution to the empirical asset pricing literature However, we do
not review it here, given the focus of this chapter.
366
E Allen and R Michaely
12 months If no dividends were paid during this period, they assumed that the expected
dividend was zero.
They repeated the second step for every month included in the period 1936 to 1977.
They estimated Ait+l by using the previous 60 months of data They provided an
updated estimate of the expected dividend yield for each stock for each one of the
test months.
This sequence of cross-sectional regressions results in a time series of a3,'s The
estimate of a 3 is the mean of this series They compute the standard error of the
estimate from the time series of the a3t's in a straightforward manner Litzenberger and
Ramaswamy (1979, 1980) found that a3 was positive and significantly different from
zero Using MLE and GLS procedures, Litzenberger and Ramaswamy corrected for the
error in variables and heteroskedasticity problems presented in the data However, the
empirical regularity they documented a positive and statistically significant dividend
yield coefficient was not sensitive to which method they used The various procedures
yielded similar estimated coefficients with minor differences in the significance level.
Litzenberger and Ramaswamy interpreted their finding as consistent with Brennan's
(1970) after-tax CAPM That is, the positive dividend yield coefficient was evidence
of a dividend tax effect.
Miller and Scholes (1982) argue that the positive yield coefficient found by
Litzenberger and Ramaswamy was not a manifestation of a tax effect, but an artifact
of two information biases First, Litzenberger and Ramaswamy's estimate of the nextmonth dividend yield could be correlated with month t information Of the firms
paying dividends, about 40 % announced and paid the dividend (i e , the ex-dividend
day) in the same month Using the Litzenberger and Ramaswamy yield definition
assumes that the ex-dividend month is known a priori even for ex-months in which
dividends were not declared in advance.
Second, Litzenberger and Ramaswamy ignored the potential effect of dividend
omission announcements An omission announcement, which is associated with bad
news, will tend to bias upward the dividend yield coefficient, since it reduces the return
of the zero yield group The effect of these informational biases is the center of the
debate between Litzenberger and Ramaswamy (1982) and Miller and Scholes (1982).
Miller and Scholes showed that when they included only dividends declared in
advance in the sample, or when they defined the dividend yield as the dividend yield in
month t 12, the yield coefficient was statistically insignificant Based on these results,
Miller and Scholes attributed the Litzenberger and Ramaswamy results to information,
rather than tax effects.
Responding to this criticism, Litzenberger and Ramaswamy ( 1982) constructed a
dividend-yield variable that incorporated only such information as investors could
possess at the time Their sample contained only stocks that either declared in
month t 1 and paid in month t, or stocks that paid in month t 1 and therefore
were not likely to repay in the current month Using the "information-free" sample,
Litzenberger and Ramaswamy found the yield coefficient was positive and significant.
Miller and Scholes remained unconvinced.
Ch 7:
Payout Policy
367
To resolve the informational issue, Kalay and Michaely (2000) performed the
Litzenberger and Ramaswamy experiment on weekly data, excluding all weeks
containing both the announcement and ex-day ( 3 4 % of the sample) They also
excluded all weeks containing dividend omission announcements Nevertheless, they
found a positive and significant yield coefficient, implying that information is not
the driving force behind the Litzenberger and Ramaswamy result The question still
remains whether the positive yield coefficient found by Litzenberger and Ramaswamy
can be attributed to taxes Kalay and Michaely (2000) argue that the single-period
model derived by Brennan (1970) and Litzenberger and Ramaswamy ( 1979) predicts
cross-sectional return variation as a function of dividend yield In contrast, the
Litzenberger and Ramaswamy test of Brennan's model is inadvertently designed to
discover whether the ex-dividend period offers unusually large risk-adjusted returns
(i.e , time-series return variation).
Litzenberger and Ramaswamy classified stocks as dividend-paying stocks only
during the ex-dividend months For example, they classify a stock that pays quarterly
dividends to the zero dividend yield group in two thirds of the months Therefore, when
Litzenberger and Ramaswamy find a significant positive dividend yield coefficient
in a Fama-Macbeth type test, it is not clear how to interpret these findings Are
their findings due to cross-sectional differences in dividend yield, which can then be
interpreted as evidence consistent with the Brennan model, or are their results evidence
of time-series variations in return between dividend-paying and non-dividend paying
months? In other words, can we conclude from the Litzenberger and Ramaswamy
results that higher-dividend-yield stocks show larger long-run (e g , annual) riskadjusted pretax returns (hereafter, cross-sectional return variations)? Or, do their results
merely point out that stocks experience higher risk-adjusted pretax returns during their
ex-month (hereafter, time-series return variations), and tell us little about the relation
between long-run pretax risk-adjusted returns and yields? Time-series return variation,
per se, is not evidence of a tax effect.
Since most stocks pay dividends quarterly, trying to avoid dividend income involves
realizing short-term capital gains Under USA tax laws, short-term capital gains are
taxed as ordinary income Thus, even though a long-term investor prefers long-term
capital gains to dividend income, he or she does not require a larger pretax riskadjusted return during only the ex-dividend period Therefore, the implications of the
Brennan model, combined with the USA tax code, is that differences in tax rates
between dividend income and long-term capital gains income should result in crosssectional return variation As do other studies (such as the ex-day studies), Kalay and
Michaely find strong evidence of time-series return variation around the ex-day period.
However, there is no evidence of cross-section return variation This result does not
support the Brennan's and Litzenberger and Ramaswamy's buy-and-hold models.
Another potential problem is whether some omitted risk factors (other than beta)
that are correlated with dividend yield, rather than taxes, can explain the positive
yield coefficient As a first indication of the potential importance of some omitted
risk factors, Miller and Scholes (1982) demonstrated that when the reciprocal of
368
E Allen and R Michaely
price, ( 1/P), is incorporated in the regression equation instead of the dividend
yield, (D/P), its coefficient is still positive and significant This issue was thoroughly
investigated by Chen, Grundy and Stambaugh (1990) Categorizing all dividend-paying
stocks into 20 portfolios according to size and yield, they found that when they used a
single risk factor, large firms with high dividend yield were the only ones to experience
a positive yield coefficient; and when they used two risk factor models, the yield
coefficient was significant for only one of the 20 portfolios.
As also suggested by Miller and Scholes (1982) and Hess ( 1983), Chen, Grundy
and Stambaugh (1990) presented evidence that dividend yield and risk measures were
cross-sectionally correlated When they allowed the risk measures to vary, they found
that the yield coefficient was positive but insignificant Chen, Grundy, and Stambaugh
showed that the positive association between yield and their portfolios' returns could
be explained by a time-varying risk premium that was correlated with yield Thus,
they concluded that there was no reliable relation between cross-sectional variation in
returns and dividend yield that is a consequence of a tax penalty.
Fama and French (1993) offer an interesting insight that is relevant to this issue They
argue that the yield coefficient might capture factors other than taxes, and that those
other factors might affect assets' returns They then show that when using the threefactor model, there is no trace of different intercepts among portfolios with different
dividend yields.
Summing up, a growing body of evidence shows that within static, single-period
equilibrium models, there is no convincing evidence of a significant cross-sectional
relation between stocks' returns and their dividend yields Perhaps a more promising
avenue for investigating this issue is to examine a model that allows for dynamic
trading around the ex-dividend day.
5.2 Dynamic models
An important development in the literature on taxes and dividends was the realization
that investors could trade dynamically to reduce their tax liability The first paper to
emphasize this aspect was that of Miller and Scholes ( 1978) They argued that there
were a number of dynamic strategies that allowed investors to avoid taxes, and that in
perfect capital markets all taxes could be avoided This observation brings us back to
the case in which dividend policy is irrelevant However, in practice, the transaction
costs of pursuing these strategies appear to be too high to make them empirically
significant.
An area where dynamic strategies appear to be more empirically relevant is trading
around the ex-date A number of studies, starting with Kalay (1982 a), have studied the
implications of this strategy We look at both types of approach.
5.2 1 Dynamic tax avoidance strategies
Miller and Scholes ( 1978) suggested an ingenious strategy for avoiding taxes By
borrowing and investing the proceeds with tax-free institutions, such as insurance
Ch 7:
Payout Policy
369
companies or pension funds, investors could create an interest deduction that allowed
them to avoid taxes Since there were assets that were held to offset the borrowing,
the position could be closed out at an appropriate point.
Several other dynamic tax avoidance strategies were suggested by Stiglitz (1983) If
individuals can easily "launder" dividends so they do not have to pay taxes on them,
then essentially, we are back in a Miller and Modigliani world, and dividend policy is
irrelevant.
However, there is little evidence that investors are actually using this or other
such strategies Peterson, Peterson and Ang (1985) showed that individual investors'
marginal tax on dividend income has been about double the marginal tax rate they
pay on capital gains income This evidence does not support a widespread use of tax
avoidance strategies of the type described by Miller and Scholes Rather, it suggests
that the transaction costs of such strategies are too high to be useful to investors.
5.2 2 Dynamic ex-dividend day strategies
Several studies have considered dynamic trading strategies around the ex-dividend
day The basic idea is that investors can change their trading patterns around the exdividend day to capture or avoid the upcoming dividend Kalay (1982 a) argued that
in a risk-neutral world, without any restrictions or imperfections such as transaction
costs, dynamic arbitrage could eliminate a tax effect in prices Traders with the same
tax rate on dividends and capital gains will buy the stock before it goes ex-dividend
and sell it just after Without risk or transaction costs, the arbitrage will ensure that
the price drop is equal to the dividend, i e ,
PB
PA
D
-D
1
( 15)
If there are transaction costs, and no price uncertainty, then (PB PA)/D must lie
within a range around one This range will be larger the greater are transaction costs.
However, Kalay (1982 a) did not explicitly account for the risk involved in the ex-day
trading.
The framework used by Michaely and Vila ( 1995) describes the ex-day price
formation within a dynamic equilibrium framework in which, because of taxes, agents
have a heterogeneous valuation of a publicly traded asset The intuition behind their
model is that an investor equates the marginal benefit of trading that arises from being
more heavily invested in the dividend-paying stock with the marginal cost that arises
from the deviation from optimal risk sharing.
Agents trade because they have heterogeneous valuation of dividends relative
to capital gains (on an after-tax basis) This framework incorporates short-term,
corporate, and individual investors' desire to trade around the ex-dividend day The
model explicitly accounts for the risk involved in the trade, and concludes that it is
not arbitrage, but equilibrium, that determines prices and volume In other words, the
370
E Allen and R Michaely
existence of risk precludes pure arbitrage opportunities and prices are determined in
equilibrium Consequently, no trader will attempt to take an unlimited position in the
stock, regardless of his or her tax preference.
The model illustrates that although two-period models like those of Brennan (1970)
or Litzenberger and Ramaswamy ( 1979) adequately describe the effect of taxes on
portfolio holdings in a static equilibrium, they mask a qualitative difference between
models of financial markets with and without taxation, namely, optimal tax-induced
trading Because of the dynamic nature of the Michaely and Vila model, it is possible
to derive volume and price behavior implications As it turns out, they can extract the
second moment of the heterogeneity distribution (i e , the dispersion in the after-tax
valuation of dividends) from the trading volume around the ex-day.
Using this framework, it is possible to show that in equilibrium, the expected
price drop in relation to the dividend reflects the average preference of all traders,
weighted by their risk tolerance and wealth, and the risk involved in the ex-dividend
day transaction:
E(Pr) =
E(PelPc)
D
_ X(o2/K)
D
where E(Pr) = is the expected price drop in relation to the dividend amount (hereafter,
"the premium"); Pc = the cum-day price; Pe = the ex-day price; D = the dividend
amount; ,2 = the ex-day variance; K = the after-tax weighted average of investors'
risk tolerance; X = the supply of securities; ai = -r = the relative tax preference
of dividend relative to capital gains; a = E'kai = the average of investors tax
preferences (ai) weighted by their risk tolerance (ki).
As it turns out, unless a perfect tax clientele exists in which different groups
hold different stocks rather than just different quantities of the same stock, it is
not possible to infer tax rates from price alone However, we can infer the crosssectional distribution of tax rates by using both price and volume data By observing
the premium alone, we can infer only the weighted-average relative tax rates, not the
entire distribution of tax rates for the trading population Michaely and Vila ( 1995)
show that the second moment of the distribution could be extracted from the volume
behavior on the ex-dividend day 7
This point can be illustrated by the following example Assume that there are three
groups of traders in the marketplace with a marginal rate of substitution between
dividends and capital gains income of 0 75, 1 0, and 1 25, respectively Assume further
that the average price drop relative to the dividend amount is one Using the standard
7 Boyd and Jagannathan (1994) provide a model in which proportional transactions costs faced by
different classes of traders induce a non-linear relationship between ex-day price movement and dividend
yield.
Ch 7:
PayoutPolicy
371
analysis, we can conclude that the second group dominates the ex-dividend day price
determination.
However, this conclusion might not be valid For example, suppose that half of the
traders are from the first group and half are from the third group, and both have the
same effect on prices This market composition will also result in a relative price drop
equal to the dividend amount The only way to distinguish between the two scenarios
is by incorporating volume into the analysis In the first case, there are no gains from
trade, and therefore no excess volume on the ex-dividend day In the second case, there
are gains from trade, excess volume is observed, and the particular equilibrium point
is at a relative price drop equal to one The model allows the researcher to distinguish
between such cases.
N
AV = ({D
(a(
)(Ki/ 2 ) },
(17)
i=l
where AV is the abnormal trading volume on the ex-dividend day.
This framework also incorporates the Elton and Gruber ( 1970) and Kalay (1982a)
analyses in Equation ( 17) Both analyses assume an arbitrage framework in the sense
that the last term in Equation (17) is zero, i e , there is no risk involved in the trade.
Elton and Gruber assume that for some exogenous reason (e g , transaction costs), the
only trade around the ex-day will be done by investors within the same tax clientele
group In other words, if there is a perfect holding clientele and all trading is done
intra-group, then the relative price drop will reflect the marginal value of dividends
relative to capital gains (Note that in this scenario, the marginal and the weighted
average values are the same) In this case there are two reasons why there will be
no abnormal trading volume around the ex-dividend day First, since all trades are
within the same clientele group, all relevant traders value the dividend equally, and
there are no gains from trade Second, there are no incentives for investors within the
clientele group to delay or accelerate trades because of the upcoming dividends as, for
example, suggested by Grundy ( 1985) In other words, Elton and Gruber suggest that
taxes affect price, but do not locally affect investors' behavior lno extra trading, as in
Equation (17)l Kalay takes the opposite view Taxes affect behavior but not prices,
i.e , through their trading the arbitrageurs will ensure that the price drop equals the
dividend amount Since Kalay uses the arbitrage framework, he can show that shortterm investors may take an unlimited position in the stock as long as the expected
price drop is not equal to the dividend amount.
Tests of these propositions have taken several forms Most studies examine the price
behavior and infer investors' preferences and behavior from prices With only a few
exceptions lGrundy ( 1985), Lakonishok and Vermaelen (1986), Michaely and Vila
( 1995, 1996), and Michaely and Murgia ( 1995)l, researchers have devoted much less
attention to a direct examination by using volume to determine the effect of differential
taxes on investors' trading behavior Researchers have almost always found that the
average price drop between the cum and the ex-day is lower than the dividend amount
372
E Allen and R Michaely
lsee Elton and Gruber (1970), Kalay (1982 a), Eades, Hess and Kim (1984), and Poterba
and Summers (1984), among othersl 8 For example, Eades, Hess and Kim (1984) find
an excess return of 0 142 % on the ex-dividend day and a cumulative excess return of
0.334 % in the ten days surrounding the ex-day (day -5 to day + 5, relative to the exdividend day) The positive abnormal return before the ex-day and the negative excess
return after the ex-day indicate that investors who prefer dividends start to accumulate
the stock several days before the event (its timing is known in advance) Likewise,
the negative return after the event supports the notion that investors' selling after the
ex-day is more gradual than we would predict in perfect markets.
Many of these studies also find that the average premium increases with dividend
yield lsee, for example, Elton and Gruber (1970), Kalay ( 1982a), Lakonishok and
Vermaelen ( 1986) and Boyd and Jagannathan (1994)l This finding is consistent
with tax clienteles (The tax clientele we allude to can be either a holding clientele
or a trading clientele Only examination of trading volume can separate the two).
Corporations, which prefer dividends over capital gains, and tax free institutions, which
are indifferent to the form of payment, hold high-yield stocks The ex-day premium
reflects those preferences Eades, Hess and Kim's (1984) findings of a premium greater
than one for preferred stock is also consistent with this idea That is, this group of
stocks pays a high dividend yield, and the dominant traders of these stocks (at least
around the ex-day) are the corporate traders, who prefer dividends.
Another way to examine the effect of taxes on ex-day price behavior is to examine
the effect of tax changes If taxes affect investors' decisions on buying or selling
stocks around the ex-day, a change in the relative taxation of dividends to capital
gains should affect prices Poterba and Summers (1984) looked at the British market
before and after tax changes and found evidence that indicated a tax effect Barclay
(1987) compared the ex-day price behavior prior to the introduction of federal taxes in
1913 with its behavior in the years 1962 to 1985 He found that the average premium
was not significantly different from one before the enactment of the federal taxes, and
significantly below one after Barclay concluded that the higher taxes on dividends
after 1913 caused investors to discount their value.
Michaely ( 1991) examined the effect of the 1986 Tax Reform Act (TRA) on exday stock price behavior The 1986 TRA eliminated the preferential tax rates for longterm capital gains that had been adopted in 1921; dividend income and realized capital
gains were taxed equally after the reform If taxes are at work, we would expect the
premium to be closer to one after the 1986 TRA (The premium is defined as the price
difference between the ex-day and the cum day, relative to the amount of dividend
paid) Surprisingly, this was not the case The average premium, both before and after
the TRA, was not lower than one Comparing his results to the Elton and Gruber
study, which used data from the 1960s, Michaely concludes that the change in the
8 For international evidence, see Kato and Loewenstein (1995) for the Japanese market, Lakonishok and
Vermaelen (1983) for the Canadian market, and Michaely and Murgia (1995) for the Italian market.
Ch 7:
373
Payout Policy
Table 8
Ex-dividend day premium a
Period
Mean
premium
SD
Z Value
% above One
Fisher test
1966-67
0 838
1 44
-7 23
46 1
-4 94
1986
1 054
1 32
2 32
49 9
-0 03
1987
1 028
1 229
1 33
50 7
0 80
1988
0 998
0 821
0 168
NA
NA
a This table presents the average premiums (price drop relative to dividend paid) for three time periods.
The first period, 1966 and 1967, is in Elton and Gruber (1970) and Kalay (1982a); the second, third, and
fourth periods, 1986, 1987, and 1988, are the periods before the implementation of the 1986 TRA, the
transition year, and after the implementation of the 1986 TRA, respectively We adjust premiums to the
overall market movements using the OLS market model Premiums are corrected for heteroskedasticity.
Results are taken from Michaely (1991) The null hypothesis is that the mean premium equals one.
relative pricing of dividends between the 1960 S and the 1980 S was not because of
taxes, but perhaps, because of the change in weights of the various trading groups.
Facing lower transaction costs in the equity, options, and futures markets, institutional
and corporate investors seem to trade more around the ex-day in the latter period.
Thus, their preferences have a greater effect on the price formation These results are
summarized in Table 8.
Although in static models, such as Brennan (1970) or Elton and Gruber (1970),
transaction costs can be safely ignored (since investors trade only once), in the dynamic
models they are potentially much more important If investors trade in and out of
stocks because of taxes, the multiple rounds of trades could result in a nontrivial cost
of transacting Disregarding risk, Kalay ( 1982a) showed that the "arbitrage" by the
short-term traders would take place as long as the level of transaction costs was low
enough Indeed, Karpoff and Walkling (1988, 1990) showed that excess returns were
lower for stocks with lower transaction costs This is especially pronounced for stocks
with high dividend yields, both on the NYSE/Amex and for Nasdaq stocks In other
words, corporations and short-term traders have a greater effect on the ex-day prices
of stocks with lower levels of transaction costs.
When the risk involved in the ex-day trading is accounted for, the effect of
transaction costs on trading is not as straightforward Michaely, Vila and Wang (1996)
developed a formal model that incorporated the effect of both transaction costs and
risk on ex-day prices and trading As expected, they predicted that transaction costs
would reduce the volume of trade.
More interesting is the interaction between transaction costs and risk First, with
or without transaction costs, risk reduces volume However, unlike price, volume is
negatively affected by the level of idiosyncratic risk As the level of transaction costs
increases, systematic risk negatively affects the volume of trade The reason is simple.
Without transaction costs, investors can afford to hedge all of the systematic risk In the
374
E Allen and R Michaely
presence of transaction costs, the systematic risk is not completely hedged; therefore
it affects the amount of trading.
Empirical evidence supports these results Grundy ( 1985), Lakonishok and Vermaelen (1986), and Michaely and Vila (1996) show that the abnormal volume on and
around the ex-day is significant This evidence indicates that a perfect tax clientele
where investors hold strictly different stocks, does not exist (In a perfect clientele,
no ex-day trading will take place, because each clientele group will strictly hold
only stocks with the dividend yield appropriate to its type) Moreover, the evidence
questions the idea that the marginal tax rate can be inferred from prices alone.
Michaely and Vila ( 1996) show that both risk and transaction costs affect volume.
They demonstrate that stocks with lower transaction costs experience higher abnormal
volume, and that the differences are substantial For example, between 1988 and 1990,
stocks with a low average bid-ask spread experienced an abnormal trading volume of
556 % compared with an abnormal trading volume of 78 % for high-spread stocks The
differences were even larger when they looked at only stocks with high dividend yields,
where the incentives to trade are larger Moreover, they find that idiosyncratic risk
significantly affects trading volume and that market risk has a greater effect (negative)
on trading volume when the level of transaction costs is higher.
Some of these effects are captured in the following regression analysis:
CAV = 1 89 + 63 17 (P)i
(15 8) (8 5)
0 49
O 37,+ O134 SIZE
(-18 2) (-9 3) (5 7),
( 18)
where CA Vj is the cumulative abnormal volume in the 11 days around the ex-dividend
day; (D/P)i is the stock's dividend yield, calculated as the dividend amount relative to
the cum-day price; i/Om is the idiosyncratic risk scaled by the market risk during
the same time period; i is the systematic risk; and SIZE is the market value of
equity, which is used as a proxy for the cross-sectional variation in transaction costs.
t-Statistics are reported in parentheses.
Both the idiosyncratic risk and the systematic risk are negative (and significant) The
idiosyncratic risk is about 35 % higher (in absolute value) than the beta risk coefficient.
The fact that both risk factors are significant indicates that investors do not hedge all of
their risk exposure If they did, the beta coefficient would have been zero The reason
for the incomplete hedging is transaction costs.
Koski and Michaely (2000) report that ex-day trading volume increases more in
orders of magnitude when traders are able to arrange the cum-day/ex-day trading using
non-standard settlement days That is, by virtually eliminating the risk exposure and
reducing transaction costs, volume increases significantly.
Koski and Michaely (2000) examine very large block trades around the ex-day.
Block trades involve a large purchase and subsequent sale of the dividend-paying stock
within minutes (with a different settlement day for each transaction) These trades are
done through bilateral bargaining between the two parties involved, usually Japanese
insurance companies on the buying side and a USA institution on the selling side.
Ch 7:
Payout Policy
375
This procedure substantially reduces the risk exposure (and transaction costs) relative
to "conventional" dividend-capture trading 9
As discussed earlier, examining prices alone may mask investors' tax preferences
and the trading motives that are related to taxes Kalay ( 1982a) and Eades, Hess and
Kim (1984), and more recently Bali and Hite ( 1998) and Frank and Jagannathan
(1998), have raised two additional obstacles in interpreting the ex-day price drop
as evidence that differential taxes affect prices and trading behavior First, that
discreteness in prices may cause a bias in measuring the ex-day price drop relative to
the dividend (Until recently, the minimum tick size was one eighth in the USA ) These
studies, and those by Dubofsky (1992) and Bali and Hite ( 1998), show that this bias
may cause the average price drop to be less than the dividend amount Second, that the
high correlation between dividend yield and the dollar amount of dividend paid (high
yield stocks tend to be stocks that pay large dividends) can also result in an association
between relative price drop and dividend yields 3/4the very same evidence that many
studies have attributed to dividend clienteles Eades, Hess and Kim (1984) and Frank
and Jagannathan (1998) present supporting evidence Frank and Jagannathan find that
the average price drop is less than the dividend in Hong Kong, where dividends and
capital gains are not taxed Eades, Hess and Kim (1984) find that the average price
drop is less than the dividend for non-taxable distributions in the USA This collective
evidence seems to indicate that institutional factors such as tick size play a role in the
determination of the ex-day prices.
However, in light of the results of other studies, the conclusion that the entire exday price anomaly is driven by the tick size is unlikely For example, Barclay (1987)
finds that prior to the introduction of the income tax in the USA, the average ex-day
price drop was equal to the dividend amount, despite the fact that even then, prices
were quoted in discrete multiples Michaely (1991) also finds that the average price
drop around the 1986 TRA was essentially equal to the dividend amount (see Table 8).
Again, also during this time period, prices were quoted in one-eighth increments.
Green and Rydqvist ( 1999) conducted an experiment relevant to this issue using data
on Swedish lottery bonds Taxes in the lottery bond market lead investors to prefer
cash to capital gains Some of the friction identified in the literature, such as price
discreteness, would work in the opposite way In addition, the activity of arbitrageurs
is not an issue Green and Rydqvist find that both the price drop around the ex-day
and volume behavior around this event reflects the relative tax advantage of the cash
distribution Their findings support the interpretation of the ex-day price behavior as
tax-motivated and that this behavior cannot be attributed to market frictions.
The information on volume behavior in the USA lLakonishok and Vermaelen
(1986), Michaely and Vila ( 1996)l and other countries such as Italy lMichaely
9 Michaely and Murgia (1995) show that the trading volume of both block trades and non-block trades
(on the Milan stock exchange) increases substantially for stocks with high dividend yield and low
transaction costs Their findings support the notion that low transaction costs enhance ex-day trading.
376
E Allen and R Michaely
and Murgia (1995)l, Japan lKato and Loewenstein (1995)l and Sweden lGreen and
Rydqvist (1999)l also clearly indicates that there is abnormal activity around the exdividend day The evidence also shows that the trading activity is positively related to
the magnitude of the dividend and negatively related to the level of transaction costs
and risk The evidence is consistent with the notion that this trading activity is related
to differential taxes.
5.3 Dividends and taxes
conclusions
Differential taxes affect both prices (at least around the ex-dividend day) and investors'
trading decisions In most periods examined, the average price drop is less than the
dividend paid, implying a negative effect on value The entire price behavior cannot be
attributed to measurement errors or market frictions However, it is also rather clear that
market imperfections such as transaction costs, the inability to fully hedge, and price
discreteness inhibit tax-motivated trading Absent these imperfections, it is possible
that no trace of the tax effect would show up in the pricing data So, while in perfect
and complete capital markets dividends may not affect value, this relation is much
less clear in incomplete markets with transaction costs The theory and some of the
empirical evidence indicate that taxes do matter, and that dividends reduce value when
risk cannot be fully hedged and transactions are costly.
Overall, the evidence from the ex-day studies appears to indicate that from a tax
perspective, dividends should be minimized The volume of trade around these events
is much higher than usual, indicating that the shares change hands from one investor'
group to the other This evidence tells us that taxes affect behavior.
The facts also indicate that a pure dividend-related tax clientele does not exist First,
there is clear evidence for intergroup ex-day trading that is motivated by taxes It is also
apparent that ex-day trading volume increases as the degree of tax heterogeneity among
investors increases This evidence suggests that as the benefits of trading increase, so
does trading volume Second, direct examination of individuals' tax returns indicates
that throughout most of the period 1973-1999, individuals in high tax brackets receive
substantial amounts of taxable dividends, which refutes the tax clientele argument.
Third, there is no evidence that dividend changes indicate any significant clientele
shift, as we would expect if dividend clienteles did exist.
One way of looking for evidence of clientele shifts is to see whether the turnover
rate for firms that initiate or omit dividends shows a marked change following the
announcement Richardson, Sefcik and Thompson (1986) do this for 192 firms that
initiated dividends They concluded that the volume response is primarily in response
to the news contained in the initiation announcement rather than to a clientele shift.
Michaely, Thaler and Womack (1995) examined the turnover of both initiating and
omitting firms They concluded that the relatively minor increase in volume around
the event and the absence of an increase in the six months thereafter was too low to
be consistent with a significant clientele shift.
Ch 7:
377
Payout Policy
Michaely, Thaler and Womack also directly investigated whether the share of
institutional ownership changed after dividend omission For the 182 firms with
available data, they found that the average institutional ownership was 30 % in the
three years prior to the omission and was 30 9 % after This evidence further supports
the impression that dividend changes do not produce dramatic changes in ownership.
However, Brav and Heaton (1998) find a drop in institutional ownership around
dividend omissions after the ERISA regulations took effect in 1974 Binay (2001)
examines both initiations and omissions and reports a significant drop in institutional
ownership after omissions and an increase in institutional ownership after initiations.
Perez-Gonzalez (2000) looks at changes in firms' dividend policy as a result of tax
reforms He finds that dividend policy is much more affected by the tax reform when
the largest shareholder is an individual than it is when the largest shareholder is an
institution or when there is no large shareholder Finally, Del Guercio ( 1996) examines
the role of dividends in the portfolio selection of institutions She finds that after
controlling for several other factors such as market capitalization, liquidity, risk, and
S&P ranking, dividend yield has no power in explaining banks' portfolio choice, and is
a negative indicator in mutual funds' portfolio choice Overall, her evidence indicates
that the prudent man rule has a role in portfolio selection but that dividends do not
play a major role in it.
In light of the above discussion, perhaps it is less surprising that tests of the static
models with taxes have not been successful These tests cannot accommodate dynamic
trading strategies, which seem to be important in this context In addition, time-varying
risk may result in spurious positive yield coefficients lChen, Grundy and Stambaugh
( 1990)l and missing pricing factors can also result in a positive yield coefficient lFama
and French ( 1993)l As Naranjo, Nimalendran and Ryngaert ( 1998) show, even when
they do find a dividend yield effect, it is difficult to attribute it to taxes, since it does not
vary with relative taxation and is absent in large-cap stocks Indeed, the ex-dividend
day studies that account for these effects have been more successful in identifying the
extent to which taxes affect prices and traders' behavior.
6 Asymmetric information and incomplete contracts
theory
6.1 Signaling and adverse selection models
Capital markets are imperfect, but not just because individuals and corporations have
to pay taxes Another potentially important imperfection relates to the information
structure: if insiders have better information about the firm's future cash flows, many
researchers suggest that dividends might convey information about the firm's prospects:
dividends might convey information not previously known to the market, or they may
be used as a costly signal to change market perceptions concerning future earnings
prospects.
Using the sources and uses of funds identity, and assuming the firm's investment
is known, dividend announcements may convey information about current earnings
378
E Allen and R Michaely
(and maybe even about future earnings, if earnings are serially correlated) even in the
absence of any signaling motive Since investment is known, dividends are then the
residual Thus, larger-than-expected dividends imply higher earnings Since the market
does not know the current level of earnings, higher-than-anticipated earnings would
lead to a positive stock price increase (When we talk about dividends in this context,
what we really mean is net dividends We define these in Section 4 as dividends plus
repurchases minus equity issues) This interpretation of dividend announcements is not
new and originated with Miller and Modigliani (1961) and later to the more formal
argument in Miller and Rock (1985).
However, it was not until the late 1970 S and early 1980S that any signaling models
were developed The best known are those of Bhattacharya ( 1979), Miller and Rock
( 1985) and John and Williams ( 1985) The basic intuitive idea in all these models
is that firms adjust dividends to signal their prospects A rise in dividends typically
signals that the firm will do better, and a decrease suggests that it will do worse These
theories may explain why firms pay out so much of their earnings as dividends Thus,
they are consistent with the first empirical observation.
However, in this context one of the central questions that arises is why firms use
dividends, and not share repurchases or some other less costly means of signaling, to
convey their prospects to investors.
Bhattacharya (1979) used a two-period model in which the firm's managers act in
the original shareholders' interests At time zero, the managers invest in a project.
The managers know the expected profitability of this investment, but investors do not.
At this time, the managers also "commit" to a dividend policy At time 1, the project
generates a payoff that is used to pay the dividends committed to at time zero A crucial
assumption of the model is that if the payoff is insufficient to cover the dividends, the
firm must resort to outside financing and incur transaction costs in doing so.
At time zero, the managers can signal that the firm's project is good by committing
to a large dividend at time 1 If a firm does indeed have a good project, it will usually
be able to pay the dividend without resorting to outside financing and therefore will not
have to bear the associated transaction costs In equilibrium, it is not worthwhile for
a firm with a bad project to do this, because it will have to resort to outside financing
more often and thus will have to bear higher transaction costs If the dividends are
high enough, these extra costs will more than offset the advantage gained from the
higher price received at time 1 Since the critical trade-off in the model is between
the transaction costs incurred by committing to a large dividend and the price paid at
time 1, it follows that similar results hold when the dividends are taxed.
Just after the dividends are paid, the firm is sold to a new group of shareholders,
which receives the payoff generated by the project at time 2 The payoffs in the two
periods are independent and identically distributed The price that the new shareholders
are prepared to pay at time 1 depends on their beliefs concerning the profitability
of the project Bhattacharya's model was a significant step forward It is consistent
with the observation that firms pay dividends even when these are taxed However,
Bhattacharya's model has been criticized on the grounds that it does not explain why
Ch 7: Payout Policy
379
firms use dividends to signal their prospects It would seem that firms could signal
better if they used share repurchases instead of dividends This way of signaling
would result in the same tradeoff between the transaction costs of resorting to outside
financing and the amount received when the firm is sold, but it would result in lower
personal taxes than when dividends are used.
Bhattacharya's model, like many dividend signaling models, has the feature that
dividends and share repurchases are perfect substitutes for one another It does not
matter whether the "good" firm signals its value through repurchasing shares or paying
dividends, because the end result will be the same: the payout increases the chances that
the firm will need outside financing that is costly Therefore, one of the implications
of these models is that dividends and repurchases are perfect substitutes, an issue we
return to in a later section.
Bhattacharya's model reveals both the strengths and weaknesses of the dividend
signaling literature Its main strength is that it is able to explain the positive market
reaction to dividend increases and to announcements of share repurchases The
explanation is based on an intuitive notion that dividends tell us something about
the firm's future prospects The model is internally consistent and assumes that both
investors and management behave in a rational manner.
However, like many such models, several of its assumptions are subject to some
criticism For example, why would a management care so much about the stock price
next period? Why is its horizon so short that it is willing to "burn money" (in the form
of a payout) just to increase the value of the firm now, especially when the true value
will be revealed next period? It is also not clear from this model why firms smooth
dividends Finally, why should a firm use dividends (or repurchases) to signal? It would
be more dramatic to burn the money in the middle of Wall Street, and it might even
be cheaper.
The dissatisfaction with early models led to the development of a number of
alternative signaling theories Miller and Rock ( 1985) also constructed a two-period
model In their model, at time zero firms invest in a project, the profitability of which
cannot be observed by investors At time 1, the project produces earnings and the firm
uses these to finance its dividend payment and its new investment Investors cannot
observe either earnings or the new level of investment An important assumption
in the Miller and Rock model is that some shareholders want to sell their holdings
in the firm at time 1, and that this factor enters managers' investment and payout
decisions.
At time 2, the firm's investments again produce earnings A critical assumption of the
model is that the firm's earnings are correlated through time This setting implies that
the firm has an incentive to make shareholders believe that the earnings at time 1 are
high so that the shareholders who sell will receive a high price Since both earnings
and investment are unobservable, a bad firm can pretend to have high earnings by
cutting its investment and paying out high dividends instead A good firm must pay
a level of dividends that is sufficiently high to make it unattractive for bad firms to
reduce their investment enough to achieve the same level.
380
E Allen and R Michaely
The Miller and Rock theory has a number of attractive features The basic story, that
firms shave investment to make dividends higher and signal high earnings, is entirely
plausible Unlike the Bhattacharya (1979) model, the Miller and rock theory does not
rely on assumptions that are difficult to interpret, such as firms being able to commit
to a dividend level.
What are its weaknesses? It is vulnerable to the standard criticism of signaling
models that we discuss above It is not clear that if taxes are introduced, dividends
remain the best form of signal It appears that share repurchases could again achieve
the same objective, but at a lower cost.
In Bhattacharya (1979), the dissipative cost that allowed signaling to occur was the
transaction cost of having to resort to outside financing In Miller and Rock (1985),
the dissipative costs arise from the distortion in the firm's investment decision John
and Williams (1985) present a model in which taxes are the dissipative cost The
theory thus meets the criticism that the same signal could be achieved at a lower
cost if the firm were to repurchase shares instead So while the Miller and Rock and
the Bhattacharya models imply that dividends and repurchases are perfect substitutes,
the John and Williams model implies that dividends and repurchases are not at all
related A firm cannot achieve its objective of higher valuation by substituting a dollar
of dividends for a dollar of capital gains.
What is the reasoning behind this result? Like other models, John and Williams's
starting point is the assumption that shareholders in a firm have liquidity needs that
they must meet by selling some of their shares The firm's managers act in the interest
of the original shareholders and know the true value of the firm Outside investors do
not If the firm is undervalued when the shareholders must meet their liquidity needs,
then these shareholders would be selling at a price below the true value However,
suppose the firm pays a dividend, which is taxed If outside investors take this as a
good signal, then the share price will rise Shareholders will have to sell less equity to
meet their liquidity needs and will maintain a higher proportionate share in the firm.
Why is it that bad firms do not find it worthwhile to imitate good ones? When
dividends are paid, it is costly to shareholders because they must pay taxes on them But
there are two benefits First, shareholders receive a higher price for the shares that are
sold Second, and more importantly, these shareholders retain a higher proportionate
share in the firm If the firm is actually undervalued, this higher proportionate share
is valuable to the shareholder If the managers' information is bad and the firm is
overvalued, the opposite is true It is this difference that allows separation If dividends
are costly enough, only firms that are actually good will benefit enough from the
higher proportionate share to make it worthwhile bearing the cost of the taxes on the
dividends.
John and Williams's model thus avoids the objection to most signaling theories of
dividends Firms do not repurchase shares to avoid taxes, because it is precisely the cost
of the taxes that makes dividends desirable This is clearly an important innovation.
What are the weaknesses of the John and Williams' theory? In terms of assumptions,
they take it as a given that shareholders must meet their liquidity needs by selling their
Ch 7:
Payout Policy
381
shares They rule out the use of debt, either by the firm or the shareholders themselves.
We could ask why the firm does not borrow and use the proceeds to repurchase its
shares Again, doing so would meet the liquidity needs of investors and would only
be worthwhile if the firm's shares were undervalued It should be possible to signal
the firm's value costlessly by repurchasing shares and thus increasing the proportionate
share held by the firm The Ross (1977) study shows that borrowing serves as a credible
signal Even if, for some reason, corporate borrowing is not possible, an alternative
is for the investors to borrow on their personal accounts instead of selling shares.
Again, this would allow them to meet their liquidity needs without incurring the cost
of signaling.
It is also not obvious that the John and Williams model's empirical implications
support dividend smoothing The best way to extend the model over a longer time is
not entirely clear If firms' prospects do not change over time, then once a firm has
signaled its type, no further dividend payments will be necessary and payouts can be
made through share repurchases If firms' prospects are constantly changing, which
seems more plausible, and if dividends signal these, we would expect that dividends
will also constantly change This prediction of the model is difficult to reconcile with
the observation that corporations smooth dividends, and in many cases do not alter
them at all for long periods of time We can also make the same criticism of the
other signaling models After the Miller and Rock ( 1985) and John and Williams
( 1985) papers, a number of other theories with multiple signals were developed.
Ambarish, John and Williams (1987) constructed a single-period model with dividends,
investment, and stock repurchases Williams ( 1988) developed a multi-period model
with these elements and showed that in the efficient signaling equilibrium, firms
typically pay dividends, choose their investments in risky assets to maximize net
present value, and issue new stock Constantinides and Grundy ( 1989) focused on
the interaction between investment decisions and repurchase and financing decisions
in a signaling equilibrium With investment fixed, a straight bond issue cannot act
as a signal, but a convertible bond issue can When investment is chosen optimally
rather than being fixed, this is no longer true; a straight bond issue can act as a
signal.
Bernheim ( 1991) also provided a theory of dividends in which signaling occurs
because dividends are taxed more heavily than repurchases In his model, the firm
controls the amount of taxes paid by varying the proportion of the total payout that
is in the form of dividends, rather than repurchases A good firm can choose the
optimal amount of taxes to provide the signal As with the John and Williams model,
Bernheim's model does not provide a good explanation of dividend smoothing.
Allen, Bernardo and Welch (2000) took a different approach to dividend signaling.
As in the previous models, dividends are a signal of good news (i e , undervaluation).
However, in their model firms pay dividends because they are interested in attracting a
better-informed clientele Untaxed institutions such as pension funds and mutual funds
are the primary holders of dividend-paying stocks because they are a tax-disadvantaged
payout method for other potential stockholders.
382
F Allen and R Michaely
Another reason for institutions to hold dividend-paying stocks is the restrictions
in institutional charters, such as the "prudent man" rules that make it more difficult
for many institutions to purchase stocks that pay either no dividends or low dividends.
According to Allen, Bernardo and Welch ( 2000), the reason good firms like institutions
to hold their stock is that these stockholders are better informed and have a relative
advantage in detecting high firm quality Low-quality firms do not have the incentive
to mimic, since they do not wish their true worth to be revealed.
Thus, taxable dividends are desirable because they allow firms' management to
signal the good quality of their firms Paying dividends increases the chance that
institutions will detect the firm's quality.
Another interesting feature of the Allen, Bernardo, and Welch model is that it does
accommodate dividend smoothing Firms that pay dividends are unlikely to reduce the
amount of the dividend, because their clientele (institutions) are precisely the kind of
investors that will punish them for it Thus, they keep dividends relatively smooth.
As in the John and Williams model, the Allen, Bernardo, and Welch model involves
a different role for dividends and repurchases They are not substitutes In fact, firms
with more asymmetric information and firms with more severe agency problems will
use dividends rather than repurchases.
Kumar ( 1988) provided a theory of dividend smoothing In his model, the managers
who make the investment decision know the true productivity type of the firm but the
outside investors do not Also, because they are less diversified the managers want to
invest less than the outside investors Managers will try to achieve lower investment
by underreporting the firm's productivity type.
Kumar shows that there cannot be a fully revealing equilibrium in which dividends
perfectly signal productivity If there were such an equilibrium, shareholders could
deduce the firm's true productivity type However, this is inconsistent with managers
underreporting.
A coarse signaling equilibrium can exist, though Within an interval of productivity,
Kumar shows that it is optimal for the different types of firm to cluster at a
corresponding dividend level This theory is consistent with smoothing, because small
changes in productivity will not usually move a firm outside the interval, so its dividend
will not change Unfortunately, this theory does not explain why share repurchases,
which are taxed less, are not used instead of dividends Kang and Kumar (1991)
have looked at the empirical relation between firm productivity and the frequency of
dividend changes Their results are consistent with Kumar's analysis.
The signaling models discussed here are important contributions They are also
intuitively appealing Firms that pay dividends, and especially firms that increase their
dividends, are firms that are undervalued by the market Thus, the most important
prediction that is common to all of these models is that dividends convey good news
about the firm's future cash flows.
The majority of the theoretical (and empirical) research has assumed that firms use
dividend changes to signal changes in future earnings or cash flows But given the less
than enthusiastic empirical endorsement this prediction has received (as we describe
Ch 7:
Payout Policy
383
in the next section), we might want to consider another possibility, that increases in
dividends convey information about changes in risk rather than about growth in future
cash flows.
By definition, the fundamental news about a firm must be about either its cash flows
or its discount rates (risk characteristics) If the good news in a dividend increase is
not about (expected) increases in future cash flow, then it might concern a decline in
(systematic) risk.
Current dividend-signaling models have very little to say about the relation between
dividend changes and risk changes Grullon, Michaely and Swaminathan (2002)
present an alternative explanation, which they refer to as the "maturity hypothesis".
They propose that there are several elements that contribute to firms becoming mature.
As firms mature, their investment opportunity set shrinks, resulting in a decline in their
future profitability But perhaps the most important consequence of a firm becoming
mature is a change in its (systematic) risk characteristics, specifically, a decline in risk.
The decline in risk most likely occurs because the firm's assets in place have become
less risky and/or the firm has fewer growth opportunities available Finally, the decline
in investment opportunities generates an increase in free cash flows, leading to an
increase in dividends Thus, a dividend increase indicates that a firm has matured.
According to the maturity hypothesis, firms increase dividends when growth
opportunities decline, which leads to a decrease in the firm's systematic risk and
profitability How, then, should the market react to a dividend increase? The dividend
increase clearly contains at least two pieces of news The good news is that the risk has
decreased, and the bad news is that profits are going to decline The positive market
reaction implies that news about risk dominates news about profitability Another
possibility is that because of agency considerations, investors treat dividend increases
as good news, in spite of the declining profitability For instance, if investors expect
managers to squander the firm's wealth by overinvesting, then a dividend increase
suggests that managers are likely to act more responsibly Thus, in addition to the good
news conveyed about a risk reduction, investors might interpret a dividend increase
as good news per se (they reduce the overinvestment problem), and the stock price
would rise Modeling the dynamic relation between firms' dividend policy, investment
opportunities, and cost of capital is still an unexplored path that could yield valuable
new insights into the determination of corporate payout policy.
6.2 Incomplete contracts
agency models
If we relax the assumption of complete (and fully enforceable) contracts, we realize
that a firm is more than just a "black box" The different forces that operate within a
firm can, at different points in time, pull it in different directions, and the interests of
different groups within a firm may conflict The three groups that are most likely to
be affected the most by a firm's dividend policy are stockholders, management, and
bondholders.
384
E Allen and R Michaely
The first conflict of interest that could affect dividend policy is between management
and stockholders As suggested by Jensen and Meckling (1976), managers of a publicly
held firm could allocate resources to activities that benefit them, but that are not in
the shareholders' best interest These activities can range from lavish expenses on
corporate jets to unjustifiable acquisitions and expansions In other words, too much
cash in the firm may result in overinvestment.
Grossman and Hart ( 1980), Easterbrook (1984) and Jensen ( 1986) have suggested
a partial solution to this problem If equityholders can minimize the cash that
management controls, they can make it much harder for management to go on
(unmonitored) spending sprees The less discretionary cash that management has, the
harder it is for them to invest in negative NPV projects One way to take unnecessary
cash from the firm is to increase the level of payout.
We note that these theories suggest a significant departure from the original Miller
and Modigliani assumption in that payout policy and investment policy are interrelated.
Paying out cash would increase firm value by reducing potential overinvestments.
Cash payouts make an appealing argument, and as we will show, it also receives
significant empirical support But payouts also have several shortcomings First, if
managers want to overinvest, either to increase their power base by acquiring more
firms, or simply to spend more on jets and hunting trips, what is the mechanism that
will force them to commit to an action that will prevent them from doing so? Or is it
the board of directors that forces them to change their payout policy? If so, what is the
information structure and the enforcement mechanism between the board of directors
and the management that allows the board to set the appropriate dividend policy exante, but not to monitor management's actions ex-post? Put another way, if the board
(which we assume is independent of the management and cares about shareholders'
best interests a very strong assumption indeed) knows that management overinvests,
why can't it monitor it better?
Several authors, most notably Zwiebel (1996), Fluck ( 1999) and Myers (2000),
address this issue in the context of capital structure, but the basic insight for
payout policy is straightforward It must be in management's self-interest to maintain
positive payout ex-post In contrast to the standard free cash flow stories, management
voluntarily commits to pay out cash because of constant potential threat of some
(limited) disciplinary actions This is also the notion that the Allen, Bernardo and
Welch (2000) paper brings to the payout policy issue Their paper highlights the role
of large outsider shareholders' constant monitoring role.
Another question asks why firms pay out in the form of dividends and not share
repurchases, since the latter are a cheaper way to take money out of management hands.
A related question is why monitor through payout and not debt? As Grossman and Hart
(1980) and Jensen ( 1986) argue, a more effective mechanism to achieve this goal is to
increase the level of debt It is harder for management to renege on a debt commitment
relative to a dividend commitment This argument can also be applied to the choice
of dividends versus repurchases If we take as given the empirical observation that
the market strongly dislikes dividend reductions and that management is therefore
Ch 7:
Payout Policy
385
reluctant to reduce dividends, then dividends represent a more effective mechanism
than repurchases to impose discipline.
Third, although the agency story offers a palatable explanation for dividend
increases, it is much less so for dividend decreases Firms increase their dividends
when they have free cash flow, and the positive market reaction to the dividend
announcement happens because the market realizes that now management will have
to be more disciplined in its action But what about dividend cuts? One possibility is
that management cuts dividends when cash flow, and hence free cash flow, has fallen.
Another possibility is that management (or the board) cuts dividends when there are
good investments, so the cut should also be greeted positively by the market Needless
to say, this does not happen In this case, the good investments could be financed by
debt.
The earlier work of Shleifer and Vishny (1986) and the more recent work by
Allen, Bernardo and Welch (2000) provides a framework that can overcome the
first two problems (management incentive to pre-commit and dividends as opposed
to repurchases) Building on the work of Grossman and Hart (1980), Shleifer and
Vishny (1986) suggested that because of conflict of interest, management should be
monitored, and this monitoring must be done by large shareholders The presence of
such shareholders increases the value of the firm because of the monitoring role they
play, and because they help facilitate takeover activities (even if they are not involved).
Thus, the board has an incentive to induce major shareholders to take a position in the
firm, especially if the firm is likely to have excess cash.
Given the favorable tax treatment of dividends by some large shareholders such
as corporations, it is possible that dividends are paid to attract this type of clientele.
Allen, Bernardo and Welch (2000) extend this analysis and show that a favorable tax
rate for institutions relative to individuals is enough for those large shareholders to
prefer dividend-paying stocks This observation is important, since now the analysis
can encompass not only corporations (as in Shleifer and Vishny), but also various types
of tax-free institutions.
This clientele will increase the value to all shareholders, including individual
shareholders, since it monitors the management and thereby increases the firm's value.
Whether indeed large shareholders are attracted to firms that pay dividends and much
less to firms that repurchase their shares is an unresolved empirical issue that is worth
pursuing 10
The second conflict of interest that may be affected by payout policy is between
stockholders and bondholders As Myers (1977) and Jensen and Meckling ( 1976)
have argued, there are some situations in which equityholders might try to expropriate
10Based on potential conflict of interest between outside shareholders and the minority shareholders
who manage the firm, Fluck (1999) presents an interesting idea in which the more effective outsiders
are in disciplining management, the more they receive in dividends Thus, the better outsiders are at
monitoring, either because of the resources they devote to it or because of their fractional ownership,
more of the profits will be distributed to shareholders.
386
E Allen and R Michaely
wealth from debtholders This wealth expropriation could come in the form of
excessive (and unanticipated) dividend payments Shareholders can reduce investments
and thereby increase dividends (investment-financed dividends), or they can raise debt
to finance the dividends (debt-financed dividends) In both cases, if debtholders do not
anticipate the shareholders' action, then the market value of debt will go down and
the market value of equity will rise.
To summarize, in this section we presented two views of why dividends are paid The
first view is that dividends convey good news The alternative view is that dividends
are in themselves good news because they resolve agency problems In the next section
we review the corresponding empirical literature.
7 Empirical evidence
7.1 Asymmetric information and signaling models
In their original paper, Miller and Modigliani suggested that if management's
expectations of future earnings affects their decisions about current dividend payouts,
then changes in dividends will convey information to the market about future earnings.
This notion has been labeled as "the information content of dividends" As discussed
earlier, this notion has been formalized in two ways: In the first, dividends are used as
an ex-ante signal of future cash flow as, for example, in Bhattacharya ( 1979) In the
second, dividends provide information about earnings as a description of the sources
and uses of funds identity as, for example, in Miller and Rock (1985) The second
alternative can be interpreted as saying that the fact that dividends convey information
does not necessarily imply that they are being used as a signal This distinction may
be subtle, but it is crucially important in interpreting the empirical tests as supporting
the signaling theory Most, if not all, of the empirical tests we are aware of cannot
help us to distinguish between these two alternatives.
The information/signaling hypotheses contain three important implications that have
been tested empirically:
(i) Dividend changes should be followed by subsequent earnings changes in the same
direction.
(ii) Unanticipated dividend changes should be accompanied by stock-price changes
in the same direction.
(iii) Unanticipated changes in dividends should be followed by revisions in the
market's expectations of future earnings in the same direction as the dividend
change.
It is important to note that all of the above implications are necessary, but not
sufficient, conditions for dividend signaling The condition that earnings changes will
follow dividend changes is the most basic If this condition is not met, we can conclude
that dividends do not have even the potential to convey information at least not about
future cash flows, let alone to signal.
Ch 7:
Payout Policy
387
Most of the empirical literature has concentrated on the second implication, that
unexpected dividends changes are associated with price changes in the same direction.
Therefore, we start our review by describing the empirical findings on the association
between dividend changes and price changes For example, Pettit (1972) showed
that a significant price increase follows announcements of dividend increases, and
a significant price drop follows announcements of dividend decreases Aharony
and Swary ( 1980) showed that these price changes hold even after they controlled
for contemporaneous earnings announcements Using a comprehensive sample of
dividend changes of at least 10% over the period 1967-1993, Grullon, Michaely and
Swaminathan ( 2002) found that the average abnormal return to dividend increases was
1.34 % (a median of 0 95 %) and the average abnormal market reaction to dividend
decreases was -3 71 % (a median of -2 05 %).
Table 9 describes some of the characteristics of firms that change their dividends.
Both dividend-increasing and decreasing firms are larger than the typical NYSE/Amex
firm During the last four decades (the sample is from 1963 to 1998), the average
dividend-increasing firm has a dividend yield of 3 74 % before the dividend increase
and the average dividend-decreasing firm has a dividend yield of 3 29 % prior to the
dividend decrease The change in dividend is greater (in absolute terms) for firms that
decrease their dividends (-44 8% compared to 31 1%), but the frequency of a decrease
is smaller (1358 compared to 6284).
Studies by Asquith and Mullins (1983) (dividend initiations), Healy and Palepu
(1988) and Michaely, Thaler and Womack (1995) (dividend initiations and omissions)
focused on extreme changes in dividend policy Their research showed that the market
reacts quite severely to those announcements The average excess return is 3 4 % for
initiation and -7 % for omissions.
It seems that the market has an asymmetric response to dividend increases
and decreases (and for initiations and omissions), which implies that lowering
dividends carries more informational content than increasing dividends, perhaps
because reductions are more unusual, or because reductions are of greater magnitude.
Michaely, Thaler and Womack (1995) examined this issue and found that when they
controlled for the change in yield, the announcement of an omission had a larger impact
on prices than did an announcement of an initiation They also reported that the effect
of a unit change in yield (say, a 1% change in yield) had a greater effect on prices
for initiations than it did for omissions The price impact may explain, to some extent,
why managers are so reluctant to cut dividends.
There seems to be general agreement that:
(1) Dividend changes are associated with changes in stock price of the same sign
around the dividend change announcement.
(2) The immediate price reaction is related to the magnitude of the dividend.
(3) The price reaction is not symmetric for increases and reductions of dividends Announcements of reductions per se have a larger price impact than announcements
of increases.
388
E Allen and R Michaely
Table 9
Firm characteristics of dividend-changing firms a,b
Mean
Std
Median
Dividend increases (6,284 obs )
CHGDIV %
CAR %
SIZE
30 1
1 34
1,185 1
29 3
4 33
3,796 1
22 2
0 95
195 9
RSIZE
81
21
9
PRICE
29 60
24 23
24 50
DY %
3 74
2 09
3 46
Dividend decreases (1358 obs )
CHGDIV %
CAR %
SIZE
-44 8
-3 71
16 4
6 89
-45 9
-2 05
757 4
2,489 4
RSIZE
77
24
PRICE
26 31
25 31
18 50
DY %
3 29
2 19
2 87
148 0
8
This table reports the firm characteristics for a sample of firms that change their cash dividends over
the period 1967-1993 To be included in the sample, the observation must satisfy the following criteria:
1) the firm's financial data is available on CRSP and Compustat; 2) the cash dividend announcement
is not accompanied by other non-dividend events; 3) only quarterly cash dividends are considered;
4) cash dividend changes that are less than 10% or greater than 500 % are excluded; 5) cash dividend
initiations and omissions are excluded; 6) the last cash dividend payment is paid within 90 days prior to
the announcement of the cash dividend change CHGDIV is the percentage change in the cash dividend
payment, CAR is the three-day cumulative NYSE/Amex value-weighted abnormal return around the
dividend announcement, SIZE is the market value of equity at the time of the announcement of the
cash dividend change, RSIZE is the size decile ranking relative to the entire sample of firms on CRSP,
PRICE is the average price, and DY is the dividend yield at the time of the announcement of the cash
dividend change.
b Source: Grullon, Michaely and Swaminathan (2002), "Are dividend changes a sign of firm maturity"?
a
Prices can tell us not only about the immediate market reaction to the dividend
change, but also how the market perceived dividend-changing firms before the dividend
change occurred and whether the market absorbed the information contained in the
dividend change It is clear that dividend-increasing firms have done well prior to the
announcement and dividend-decreasing firms have not done as well For example, for
the period 1947-1967 Charest ( 1978) found an abnormal performance of around 4 %
in the year prior to the dividend increase month and a negative 12% for the dividend
decreasing firms Benartzi, Michaely and Thaler ( 1997) documented an average 8 6 %
abnormal return in the year prior to a dividend increase and -28 % for firms that
decreased dividends For dividend initiations and omissions, the magnitude of the
Ch 7:
Payout Policy
389
pre-announcement price movement was even more pronounced lMichaely, Thaler and
Womack (1995)l.
What is perhaps more interesting and important, from both the corporate finance
and the market efficiency perspectives, is the post-dividend-change performance.
Charest (1978) found a 4 % abnormal return in the two years after dividend increase
announcements and a negative 8% for dividend-decreasing firms Using the FamaFrench three-factor model Grullon, Michaely and Swaminathan (2002) reported a
three-year abnormal return of 8 3 % for dividend increases, which is significant.
They did not detect any abnormal performance for dividend-decreasing firms Not
surprisingly, the post-dividend abnormal performance was even more pronounced for
initiations and omissions Michaely, Thaler and Womack ( 1995) reported a marketadjusted return of almost 25 % in the three years after initiations and a negative
abnormal return of 15% in the three years after omissions.
The post-dividend announcement drift is both encouraging and disturbing from
the signaling-theory perspective It is encouraging because it is consistent with the
implication that dividend changes have some useful informational content It is
disturbing because it implies that even if firms try to signal through dividends, the
market does not "get it" or at least it does not get the full extent of the signal.
Otherwise, the entire price reaction would have happened right after the announcement.
The fact that the market doesn't get it (better future earnings or cash flows) is
problematic, since the models described above rely on the rationality assumption.
Investors and firms use the information at their disposal in the best possible way The
long-term drift does not support this assumption In other words, if investors do not
understand the signal, there is no incentive for those firms to use a costly signal.
Our next step is to examine the fundamental implication of the signaling models that dividend changes and future earnings changes move in the same direction Watts
(1973) was among the first to test the proposition that the knowledge of current
dividends improves the predictions of future earnings, over and above knowledge of
current and past earnings Using 310 firms with complete dividends and earnings
information for the years 1946-67, and annual definitions of dividends and earnings,
Watts tested whether earnings in year t + 1 could be explained by the current (year t)
and past (year t 1) levels of dividend and earnings For each firm in the sample, Watts
estimated the current and past dividend coefficients (while controlling for earnings).
Although he found that the average dividend coefficients across firms were positive,
the average t-statistic was very low In fact, only the top 10 % of the coefficients were
marginally significant Using changes in levels yielded similar results He concluded
that: "
in general, if there is any information in dividends, it is very small".
Gonedes (1978) reached a similar conclusion Penman ( 1983) also finds that after
controlling for management's future earnings forecast, there was not much information
conveyed by dividend changes themselves Interestingly, Penman also reports that many
firms with improved future earnings did not adjust their dividends accordingly.
Somewhat more in line with the theory are Healy and Palepu's (1988) results For
their sample of 131 firms that initiated dividend payments, earnings had increased
390
E Allen and R Michaely
rapidly in the past and continued to increase for the following two years However,
for their sample of 172 firms that omitted a dividend payment, the results were the
opposite of what signaling theory predicts Earnings declined in the year in which
the omission announcement took place, but then improved significantly in the next
several years For a sample of 35 firms that increased their dividends by more than
20 %, Brickley (1983) found a significant earnings increase in the year of and the year
after the dividend increase.
Perhaps we can attribute the somewhat mixed results on the relation between current
changes in dividends and future changes in earnings to the limited number of firms
used in most of these studies Another factor that makes the task difficult is knowing
how to model unexpected earnings.
Using a large number of firms and events over the period 1979-1991 and several
definitions of earnings innovations, Benartzi, Michaely and Thaler (1997) investigate
the relation between dividend changes and future changes in earnings They measure
earnings changes relative to the industry average changes in earnings that they adjusted
for earnings momentum and for mean reversion in earnings Two robust results emerge.
First, there is a very strong lagged and contemporaneous correlation between dividend
changes and earnings changes When dividends are increased earnings have gone
up There is no evidence of a positive relation between dividend changes and future
earnings changes In the two years following the dividend increase, earnings changes
were unrelated to the sign and magnitude of the dividend change.
The results were strong but perverse for dividend decreases Like Healy and Palepu
(1988), Benartzi, Michaely and Thaler (1997) find a clear pattern of earnings increase
in the two years following the dividend cut Using a sample of firms that changed their
dividends by more than 10 %, Grullon, Michaely and Swaminathan (2002) confirmed
these results They show that not only do future earnings not continue to increase, but
that the level of firms' profitability decreases in the years following announcement of
dividend increases Figure 4 presents these results The figure shows that firms move
from a period of increasing ROA before the dividend increase to a period of declining
ROA after the dividend increase.
Nissim and Ziv (2001) offer yet another look at this problem They attempt to
explain future innovation in earnings by the change in dividend, like Benartzi, Michaely
and Thaler (1997) They argue that a good control for mean reversion is the ratio of
earnings to the book value of equity (ROE) and add it as an additional explanatory
variable They advocate the inclusion of ROE to improve the model of expected
earnings, and to fix what they call an "omitted correlated variables" Rather than
adopting the natural convention of assigning a dividend change to the year in which
it actually takes place, Nissim and Ziv change this convention by assigning dividend
changes that occur in the first quarter of year t + I to year t Since we know that
dividends are very good predictor of past and current earnings, this change is bound
to strengthen the association between dividend changes and earnings growth in year 1.
Indeed using this methodology, the dividend coefficient is significant in about 50 %
of the cases when next year's earning is the dependent variable When using the
Ch 7:
Payout Policy
391
Fig 4 Level of return of assets This figure depicts the level of return on assets (ROA) based on operating
income before depreciation (Compustat annual item 13) for a sample of firms that change their dividends
over the period 1967-1993 Year O is the year in which the dividend change was announced The data
have been winsorized at the first and 99th percentiles lGrullon, Michaely and Swaminathan (2002),
"Are dividend changes a sign of firm maturity"?l
more conventional methodology, it is significant in only 25 % of the years When
using several independent variables in addition to ROE, Benartzi, Michaely and Thaler
(1997) do not find any significant relation between current changes in dividends and
future changes in earnings.
Using the Fama and French (2000) modified partial adjustment model to control for
the predictable component of future earnings changes based on lagged earnings levels
and changes, Grullon, Michaely, Benartzi and Thaler ( 2003) re-examine the relation
392
E Allen and R Michaely
between dividends and earnings changes Fama and French explicitly model the timeseries of earnings in a way that captures the empirical fact that earnings changes are
more mean-reverting in the tails They show that their model explains the evolution
of earnings much better than a model with a uniform rate of mean reversion We have
thus adopted their methods to investigate this problem ' The model is the following:
ET-ET
=
O + fl RADIVo
Bl
+ (yl + y2 NDFEDo + y3 NDFED*DF Eo + y4 PDFEDDF Eo)* DF Eo
+(
+ A2 NCEDo + A3NCED*C Eo + A4PCED*C Eo)* CEo + Er.
( 19)
In this equation DFEo is equal to RO Eo ElROEol, where ElRO Eol is the fitted value
from the cross-sectional regression of RO Eo on the log of total assets in year -1, the
market-to-book ratio of equity in year -1, and ROE 1 CEo is equal to (Eo -E_)/B 1.
NDFE Do (PDFE Do) is a dummy variable that takes the value of 1 if DF Eo is negative
(positive) and 0 otherwise, and NCE Do (PCE Do) is a dummy variable that takes the
value of I if CEo is negative (positive) and 0 otherwise As discussed in Fama and
French ( 2000), the dummy variables and squared terms in Equation 19 are included
to capture the fact that large changes in earnings revert faster than small changes and
that negative changes revert faster than positive changes It is important to control for
these non-linearities in the behavior of earnings because assuming linearity when the
true functional form is non-linear has the same consequences as leaving out relevant
independent variables.
The Grullon, Michaely, Benartzi and Thaler (2003) estimation of Equation 19 is
presented in Table 10 They find no evidence that dividend changes contain information
about future earnings growth The coefficient for RADIV is not statistically different
from zero when either year 1 earnings changes or year 2 earnings changes are the
dependent variables Furthermore, even for predictions of first year earnings growth,
the coefficient for the dividend change is significant at the 10% level in only 4 out of
the 34 years of the sample For year 2 earnings it is significantly positive at the 10%
level in just 5 out of the 34 years As documented in previous studies, this evidence
suggests that dividend changes are very unreliable predictors of future earnings.
The overall accumulated evidence does not support the assertion that dividend
changes convey information about future earnings Miller (1987) summarized the
empirical findings this way: "
dividends are better described as lagging earnings
than as leading earnings" Maybe, as Miller and Rock ( 1985) suggested, dividends
convey information about current earnings through the sources and uses of funds
identity, not because of signaling At the minimum, the empirical findings on the longterm price drift and the lack of positive association between dividend changes and
future changes in earnings raise serious questions about the validity of the dividend
11 See Fama and French (2000) for a detailed discussion of this econometric model.
Ch 7:
PayoutPolicy
393
Table 10
Regressions of raw earnings changes on dividend changes using the Fama and French approach to predict
expected earnings (see Equation 19)a,b
Year
r=
Af
Mean
0 005
T-statistic
0 56
% of t( 3j) > 1 65
r =2
22 5 %
11 8 %
Mean
0 011
T-statistic
1 13
% of t(A) > 1 65
Adjusted-R2
9 7%
12 1 %
This table reports estimates of regressions relating raw-earnings changes to dividend changes We
use the Fama-Mac Beth procedure to estimate the regression coefficients In the first stage, we estimate
cross-sectional regression coefficients each year using all the observations in that year In the secondstage, we compute time-series means and t-statistics of the cross-sectional regression coefficients The
t-statistics are adjusted for autocorrelation in the slope coefficients and reported in parentheses a, b,
and c denote significantly different from zero at the 1%, 5%, and 10% level, respectively.
b Source: Grullon, Michaely, Benartzi and Thaler (2003), "Dividend changes do not signal changes in
future profitability".
a
signaling models If firms are sending a signal through dividends, it is not a signal
about future growth in earnings or cash flows, and the market doesn't get the message.
Why would firms waste money by paying a costly dividend to send a signal that
investors do not receive?
In an interesting paper, DeAngelo, De Angelo and Skinner (1996) examined
145 firms whose annual earnings growth declined in year zero, after at least nine
years of consecutive earnings growth Thus, year zero represented the first earnings
decline in many years Their test focused on the year zero dividend decision, which
could have conveyed a lot of information to outsiders by helping the market to assess
whether the decline in earnings was permanent or transitory De Angelo, DeAngelo and
Skinner found no evidence that favorable dividend decisions (i e , dividend increases)
represented a reliable signal of superior future earnings performance There was no
evidence of positive future earnings surprises (and even some indications of negative
earnings surprises) for the 99 firms that increased their dividends Not only did the
dividend-increasing firms not experience positive earnings surprises in subsequent
years in absolute terms, their earnings performance was no better than those firms
that did not change their dividend Overall, there was no evidence that dividends had
provided a useful signal about future earnings.
None of us know for sure what market expectations are, either about prices or about
earnings But in the case of earnings, we can test for changes in market expectations
by looking at the earnings estimates of Wall Street's analysts This is how we can test
the third implication of the information/signaling theories, that unanticipated changes
394
F Allen and R Michaely
in dividends should be followed by revisions in the market's expectations of future
earnings in the same direction as the dividend change Ofer and Siegel ( 1987) used
781 dividend change events to examine how analysts change their forecast about the
current year earnings in response to the dividend changes Consistent with the positive
association between dividend changes and actual changes in concurrent year earnings
(the year of the dividend change), Ofer and Siegel found that analysts revised their
current year earnings forecast by an amount that was positively related to the size of
the announced dividend change They also provided evidence that their revision was
positively correlated with the market reaction to the announced dividend.
Most of the empirical research centers on the necessary conditions (price reaction,
subsequent earnings and changes in earnings expectations) for dividend signaling.
The outcome, as we have shown, is not encouraging Several papers looked at the
sufficient conditions for dividend signaling, most notably at taxes Recall that taxbased dividend signaling theories are based on the idea that dividends are more costly
than repurchases, and that managers intentionally use this costly device to signal
information to the market.
Bernheim and Wantz (1995) investigated the market reaction to dividend changes
during different tax regimes In periods when the relative taxes on dividends are higher
than taxes on capital gains, the signaling hypothesis implies that the market reaction to
dividend increases should be stronger, because it is more costly to pay dividends Since
it is more expensive to signal, the signals are more revealing for those who choose to
use them The free cash flow hypothesis makes the opposite prediction Since it is more
expensive to pay dividends and the benefit presumably does not change, when the taxes
on dividends are relatively higher, the market should react less favorably to dividend
increases Bernheim and Wantz's results are consistent with the dividend-signaling
hypothesis In periods of higher relative taxes on dividends, the market reaction to
dividend payments is more favorable.
However, applying nonparametric techniques that account for the nonlinear properties common to many of the dividend-signaling models in an experiment similar to
Bernheim and Wantz (1995), Bernhardt, Robertson and Farrow (1994) did not find
evidence to support the tax-based signaling models Furthermore, using data from six
years before and six years after the Tax Reform Act of 1986, Grullon and Michaely
(2001) found that the market responded much more positively to dividend increases
when dividend taxation was lower (after the tax change), a finding that is inconsistent
with tax-based signaling theories.
Amihud and Murgia (1997) examine dividend policy in Germany, where dividends
are not tax disadvantaged and in fact dividend taxation is lower than capital gains
taxation for most classes of investors In this setting, the tax-based models lsuch as
John and Williams (1985), Bernheim (1991) and Allen, Bernardo and Welch (2000)l
predict that dividend changes should not have any informational value Thus, we
should not observe a price reaction around changes in dividends However, Amihud
and Murgia (1997) find that dividend changes in Germany generated a stock price
Ch 7:
395
Payout Policy
Table 11
Average stock price before and after the dividend-increase announcement, the change in the firm cost of
capital (using the Fama-French three-factor model), the change in the average dividend payment, and
the implied change in growth; the implied change in growth is imputed from the Gordon growth model
Before the
dividend change
After the
dividend change
Actual average share
prices
$29 6
$30
Discount rates
13 2 %
12 2 %
$1 1 (Table 1)
$1 4
9 48 %
7 48 %
Average dividend
Implied growth rate
Comments
We calculate the price of $30
based on an average market
reaction of 1 43%)
We calculate the discount rate
based on Fama-French 3-factor
models and a riskless rate of 5
The average increase in
dividend is 30 % (Table 1)
reaction that was very similar to what other researchers have found in the USA This
finding is not consistent with the theory.
Grullon, Michaely and Swaminathan ( 2002) examined the relation between changes
in dividend policy and changes in the risk and growth characteristics of the firm.
Their sample comprised 7642 dividend changes announced between 1968 and 1993.
Using the Fama-French three-factor model or the CAPM, they found that firms that
increased dividends experienced a significant decline in their systematic risk, but
firms that decreased dividends experienced a significant increase in systematic risk.
Firms that increased dividends also experienced a significant decline in their return on
assets, which indicates a decline in systematic risk Capital expenditures of firms that
increased dividends stayed the same and the levels of cash and short-term investments
on their balance sheets declined.
Moreover, Grullon, Michaely and Swaminathan found that the greater the subsequent decline in risk, the more positive was the market reaction to the announced
dividend Thus, changes in risk, conditional on changes in profitability, begin to provide
an explanation for the price reaction to dividend announcements.
Using the Gordon growth model and the actual changes in risk and dividends,
Table 11 illustrates the relations between the risk reduction, the reduction in growth,
and the price reaction to the announced dividend The table shows that the average
stock price prior to the announcement is $29 6, and the average market reaction
is 1 34%, implying a post-announcement price of $30 Grullon, Michaely and
Swaminathan (2002) further reported a decline in the equity cost of capital from an
average of 13 2 % in the years before the dividend change to 12 2 % in the years after the
dividend change Now, using the Gordon growth model, we can calculate the implied
396
E Allen and R Michaely
change in growth We find that because of the decline in risk, a growth rate decline of
even 20% (from 9 48 % to 7 48 %) is still consistent with a positive market reaction.
In summary, the empirical evidence provides a strong primafacie case against the
traditional dividend signaling models First, the relation between dividend changes and
subsequent earnings changes are the opposite of what the theory predicts, so if firms
signal, the signal is not about future growth in earnings or cash flows Second, the
market doesn't "get" the signal There is a significant price drift in subsequent years.
(However, there is a change in the dividend-changing firms' risk profile, and that the
change is related both to the dividend and to subsequent performance ) Third, a crosssectional examination strongly indicates that it is the large and profitable firms and
those firms with less information asymmetries that pay the vast majority of dividends.
7.2 Agency models
Since most agency models are not as structured as the signaling models, it is difficult
to derive precise empirical implications According to the free-cash-flow models what
should happen to earnings after a dividend increase? The answer is ambiguous If
the board of directors decides to increase the dividend after management has already
invested in some negative NPV projects, then, since the payment of dividends prevents
management from continuing to invest in "bad" projects, we should expect earnings
and profitability to increase However, if the board decides on dividends before
management has the chance to overinvest, then it is difficult to say how future earnings
will be relative to past earnings If dividends increase around the time the firms face
declines in investment opportunities, then even a decline in profitability is consistent
with the free-cash-flow hypothesis.
A clearer implication of the free-cash-flow hypothesis is that the overinvestment
problem is likely to be more pronounced in stable, cash-rich companies in mature
industries without many growth opportunities Lang and Litzenberger (1989) exploited
this feature to test the free-cash-flow hypothesis, and to contrast it with the informationsignaling hypothesis The basic idea is that, according to the free cash flow hypothesis,
an increase in dividends should have a greater (positive) price impact for firms that
overinvest than for firms that do not Empirically, they identified overinvesting firms
as ones with Tobin's Q less than unity When they examined only dividend changes
that were greater than 10% (in absolute value), they found that for dividend-increase
announcements, firms with Q less than one experienced a larger price appreciation
than firms with Q greater than one For dividend-decrease announcements, firms with
Q lower than one showed a more dramatic price drop The greater effect (in absolute
value) of dividend changes on firms with lower Q is consistent with the free-cashflow hypothesis On the other hand, the information-signaling hypothesis would have
predicted a symmetric effect regardless of the ratio of market value to replacement
value.
Yoon and Starks (1995) repeated the Lang and Litzenberger experiment over a longer
time period They found that the reaction to dividend decreases was the same for high
Ch 7:
Payout Policy
397
and low Tobin's-Q firms The fact that the market reacts negatively to dividend decrease
announcements by the value-maximizing (high Q) firms is not consistent with the freecash-flow hypothesis.
Like Lang and Litzenberger (1989), Yoon and Starks found a differential reaction
to announcements of divided increases However, when they controlled for other
factors, such as the level of dividend yield, firm size, and the magnitude of the
change in the dividend yield (through a regression analysis), Yoon and Starks found a
symmetric reaction to dividend changes (both increases and decreases) between high
and low Tobin's Q firms Again, this evidence is not consistent with the free-cash-flow
hypotheses.
Grullon, Michaely and Swaminathan's findings of declining return on assets, cash
levels, and capital expenditures in the years after large dividend increases suggest
that firms that anticipate a declining investment opportunity set are the ones that are
likely to increase dividends This is consistent with the free-cash-flow hypothesis.
Lie (2000) thoroughly investigated the relation between excess funds and firms'
payout policies and found that dividend-increasing (or repurchase) firms had cash
in excess of peer firms in their industry He also showed that the market reaction
to the announcement of special dividends (and repurchases) was positively related
to the firm's amount of excess cash and negatively related to the firm's investment
opportunity set as measured by Tobin's Q These results are consistent with the idea
that limiting potential overinvestment through cash distribution, especially for firms
that have limited investment opportunities, enhances shareholder wealth.
Christie and Nanda ( 1994) examined the reaction of stocks to President Roosevelt's
unexpected announcement in 1936 of taxes on undistributed corporate profits The new
tax increased the attractiveness of dividends relative to retained earnings According
to the free-cash-flow hypothesis, firms would now have had more incentive to reduce
retained earnings and thereby reduce potential overinvestment problems, since it had
become less expensive (in relative terms) to dispense of those cash flows This effect
would have been particularly pronounced for firms that were more susceptible to
agency costs Christie and Nanda (1994) found that share prices rose in response to
the announcement of the tax change, consistent with the notion that paying dividends
may alleviate some free cash flow problems They also found that firms that were more
likely to suffer from free cash flow problems experienced a more positive price reaction
to the announcement.
The ability to monitor and the rights of outside shareholders differs across countries,
and by implication the potential severity of conflicts of interests will also differ.
La Porta, Lopez-De Silanes, Shleifer and Vishny (2000) examined the relation
between investors' protection and dividend policy across 33 countries They tested
two hypotheses The first was that when investors were better able to monitor and
enforce their objectives on management (countries with higher investors' protection),
they would also pressure management to disgorge more cash The second hypothesis
was that because of market forces (e g , management wants to maintain the ability
to raise more cash in the capital markets or wants to maintain a high stock price
398
E Allen and R Michaely
for other reasons), management would actually pay high dividends in those countries
where investors' protection was not high 12
La Porta et al (2000) found that firms in countries with better investor protection
made higher dividend payouts than did firms in countries with lower investor
protection Moreover, in countries with more legal protection, high-growth firms had
lower payout ratios This finding supports the idea that investors use their legal power
to force dividends when growth prospects are low That is, an effective legal system
provides investors with the opportunity to reduce agency costs by forcing managers to
pay out cash There is no support for the notion that managers have the incentive to
"do it on their own".
The results of La Porta et al (2000) indicate that without enforcement, management
does not have a strong incentive to "convey its quality" through payout policy There
is also no evidence that in countries with low investor protection, management will
voluntarily commit itself to pay out higher dividends and to be monitored more
frequently by the market.
Before concluding this section we discuss the empirical evidence on the relation
between the potential shareholder-debtholder conflict of interest and dividend policy.
Handjinicolaou and Kalay (1984) examined the effect of dividend-change announcements on both bond and equities prices If dividend changes are driven by
equityholders' desire to extract wealth from debtholders, then an increase in dividends
should have a positive impact on stock prices (which we know it does), and a
negative impact on bond prices The reverse should be true for dividend decreases The
alternative hypothesis, that dividends are a consequence of asymmetric information or
that they resolve free cash flow problems, implies that bond prices should move in
the same direction as equity prices Handjinicolaou and Kalay found that bond prices
dropped significantly at the announcement of dividend decreases, and did not change
significantly at dividend-increase announcements These results do not lend support to
the wealth expropriation hypothesis 13
Myers ( 1977) and Jensen and Meckling (1976) suggested that both equityholders
and bondholders may a priori agree on restricting dividends Indeed, most bond
covenants contain constraints that limit both investment and debt-financed dividends.
Kalay ( 1982b) examined these constraints and found that firms held significantly
more cash (or cash equivalents) than the minimum they needed to hold, according to
the bond covenants We can interpret Kalay's finding as a reverse wealth transfer That
is, if debt were priced under the assumption that only the minimum cash would be
12 The notion that in countries where investors' protection is low, firms would pay higher dividends
is also consistent with many of the signaling models In countries with low protection, the degree of
asymmetric information is likely to be higher, and hence the desire to pay dividends by high-quality
firms should be higher as well.
13 The asymmetry in the bond price reaction may be explained by several factors Among them is the
fact that dividend decreases are larger in absolute value than dividend increases, and therefore have a
more significant impact on both bond and stock prices.
Ch 7: Payout Policy
399
held by the corporation, then a positive reservoir would increase the market value of
debt at the expense of equityholders.
In hindsight, this is not too surprising We should not expect that large, established
firms, which are likely to have to come back to the well and seek more debt financing
at some point in the future, are going to relinquish their reputation for a small gain
at the expense of bondholders We can readily see how a one-time wealth transfer
from existing bondholders to equityholders may result in a long-term loss because
of the increase in the cost of capital When would the problem arise? In precisely
those cases where there is a great probability that the firm's time horizon is short,
e.g , the firm is in financial distress, or is about to be taken private De Angelo and
De Angelo ( 1990) found evidence that was consistent with this assertion They showed
that firms in financial distress were reluctant to cut their dividends In these cases,
not cutting dividends may constitute a significant wealth transfer from debtholders to
equityholders This is still an open question that is worth further consideration 14
8 Transaction costs and other explanations
Under certain circumstances, it is possible that investors would prefer dividends despite
the tax disadvantage of dividends relative to capital gains.
The first explanation of why firms pay dividends has to do with the "prudent man"
laws These laws and regulations are intended to protect small investors from agents
(pension funds, for example) that do not invest in their interest Private trusts, acting
under the Prudent Man Investment Act, are the most constrained fiduciaries Pension
funds are governed by the ERISA, which is less restrictive than the Prudent Man Rule.
Lastly, mutual funds are supervised by the SEC according to the Investment Company
Act of 1940, which is less restrictive than either the common law (for bank trusts) or
ERISA (for pension funds) lSee Del Guercio (1996) for information about the various
laws and regulation described herel.
Del Guercio (1996) presented evidence indicating that the Prudent Man Rule
affects investment decisions Bank managers significantly tilt the composition of their
portfolios that are viewed by the courts as being subject to the Prudent Man Rule.
Mutual funds do not Bank trusts weight their portfolios towards S&P stocks and
towards stocks that are ranked A+ (the highest ranking based on earnings and dividend
history) Mutual funds load their portfolios the other way, towards lower rank stocks.
We find it interesting that there is no difference between the portfolios' composition
of bank trusts (mainly trusts of wealthy individuals, which are highly taxed) and bank
pension funds (nontaxable entities) Both types of portfolio are weighted more towards
S&P stocks and on stocks that are ranked A+.
14 De Angelo and De Angelo (1990) allude to another link between conflict of interest and dividend
policy They report that some dividend reductions are intended to enhance the firm's bargaining position
regarding labor negotiations.
400
E Allen and R Michaely
Del Guercio went a step further Using a regression analysis, she examined the role
of dividends in the portfolio selection of institutions and found that after controlling for
several other factors, such as market capitalization, liquidity, risk, and S&P ranking,
dividend yield had no power to explain banks' portfolio choices, and had negative
explanatory power in mutual funds portfolio choice.
Overall, the evidence indicates that the Prudent Man Rule has a role in portfolio
selection, but that dividends do not play a major role (if any) This evidence is also
consistent with the information presented in Table 2, which indicates that dividend
taxation is not an issue in portfolio selection, not even for highly taxed investors.
A second motive for paying dividends is based on a transaction costs argument.
If investors want a steady flow of income from their capital investment (say, for
consumption reasons), then it is possible that dividend payments would be the cheapest
way to achieve this goal This result may hold if the cost of the alternative (i e , to sell
a portion of the holdings and receive capital gains) involves nontrivial costs These
costs might be the actual transaction costs for selling the shares, which can be quite
high for retail investors, or they could represent the time and effort spent on these
transactions.
However, this argument does not seem to be supported by the time-series evidence
on transaction costs, nor by stock ownership First, through the years, and especially
after the switch to negotiated commissions in May 1975, the transaction costs of buying
and selling shares have been substantially reduced This reduction should have resulted
in lower demand for dividends, as the alternative became cheaper The evidence in
Table 1 does not support this prediction We do not observe a reduction in dividend
payments that is related to the change in transaction costs.
Second, this argument particularly applies to individual small investors who do not
hold many shares Hence, the cost of transacting may be higher But the role of small
investors in the market place has been shrinking The overall level of dividends in the
economy has not been reduced accordingly.
Third, if this effect is in fact substantial, it should lead to an optimal dividend
policy at the aggregate level However, as Black and Scholes (1974) argued, firms
will adjust their dividend policy such that the demand for dividends by this clientele
is fulfilled Thus, in equilibrium, any specific firm should be indifferent to dividend
policy So, while this explanation can account for positive payouts despite the adverse
tax consequences, it cannot explain why, in equilibrium, firms care about the level of
dividends paid.
Shefrin and Statman (1984) suggested a third explanation as to why investors
may prefer dividend-paying stocks Rather than developing an economic model based
on maximizing behavior, they eliminated the maximizing assumptions that are the
cornerstone of neoclassical economics, and which we have maintained throughout.
Instead, Shefrin and Statman developed a theory of dividends based on several recent
theories of investors' behavior The basic idea is that even if the eventual cash received
is the same, there is a significant difference in whether it comes in the form of
Ch 7:
Payout Policy
401
dividends or as share repurchases In other words, the form of cash flow is important
for psychological reasons.
We illustrate Shefrin and Statman's approach with one of the theories they develop,
based on Thaler and Shefrin's (1981) theory of self-control Thaler and Shefrin
suggested that people have difficulties behaving rationally when they want to do
something but have problems doing so Examples that illustrate this suggestion are
the prevalence of smoking clinics, credit counselors, diet clubs, and substance-abuse
groups Individuals wish to deny themselves a present indulgence, but find that they
yield to temptation Thaler and Shefrin represented this conflict in a principal-agent
form The principal is the individual's internal planner, which expresses consistent
long-run preferences However, the responsibility for carrying out the individual's
action lies not with the planner, but with the doer, the agent.
There are two ways the planner can control the agent The first is will power The
problem is that this causes disutility The second is to avoid situations in which will
power must be used This avoidance is accomplished by adopting rules of behavior
that make it unnecessary for people to question what they are doing most of the
time.
Shefrin and Statman suggested that by having money in the form of dividends rather
than capital gains, people avoid having to make decisions about how much to consume.
Thus, they avoid letting the agent in them behave opportunistically They postulated
that the benefit of doing this was sufficient to offset the taxes on dividends.
As with the transaction costs story, the self-control explanation can account for an
aggregate positive payout policy, but not for an individual firm optimal payout policy.
That is, in equilibrium, firms will adjust their dividend policy such that the marginal
firm is indifferent to the level of dividend paid out Thus, neither the transaction costs
explanation nor the behavioral explanation can account for the positive price reaction to
dividend increases and the negative price reaction to dividend decreases Nevertheless,
this explanation is innovative and intriguing.
We also note that this explanation relies heavily on the effect that individual investors
have on market prices The need for a steady stream of cash flows combined with
significant transaction costs (the transaction costs story) may adequately describe the
actions of small retail investors, but may not hold when applied to corporate and
institutional investors Likewise, using self-control as an explanation for why firms
pay dividends is more persuasive when individual investors are the dominant force in
the marketplace As the evidence in Table 1 indicates, the level of dividend payout
did not decrease through time This evidence does not support the self-control and
transaction costs explanations.
However, Long's ( 1978) study of Citizens Utilities (CU) is illuminating CU stocks
are an almost perfect medium for examining the effect of dividend policy on prices.
The reason is that from 1955 until 1989, this company had two types of common stocks
that differed only in their dividend policy Series A stock paid a stock dividend and
402
E Allen and R Michaely
Series B stock paid a cash dividend 15 The company's charter required that the stock
dividend on Series A stock be of equal value with Series B cash dividends However,
in practice, the board of directors chose stock dividends that averaged 10 % higher than
the cash dividends Even without taxes, we would expect the price ratio of Series A
stock to Series B stock to be equal to the dividend ratio, i e , to 1 1 Long found that
the price ratio was consistently below 1 1 in the period considered This price ratio
implies a preference for cash dividends over stock dividends despite the tax penalty.
Poterba ( 1986) revisited the Citizens Utilities case For the period 1976-84, he found
that the price ratio and the dividend ratios were comparable: the average price ratio
was 1 134 and the average dividend ratio was 1 122 This evidence implies indifference
between dividend and capital gains income Poterba also examined the ex-dividend day
behavior of CU for the period 1965-84, and found that the average ex-day price decline
was less than the dividend payment This evidence supports the ex-dividend day studies
discussed previously It is hard to reconcile the ex-day evidence of the CU stocks with
the relative prices of the two stocks on ordinary days.
Hubbard and Michaely ( 1997) examined the relative prices of these two stocks
after the passage of the 1986 TRA Because the 1986 TRA substantially reduced the
advantage of receiving stock dividends rather than cash dividends, they hypothesized
that the price ratio should decrease Indeed, they found that during 1986, the price
ratio was considerably lower than in the previous years However, in the years 1987
through 1989, the price ratio rose and stayed consistently above the dividend ratio.
It seems that the evidence from the price behavior of Citizens Utilities deepens
the dividend mystery, rather than enlightening us It is difficult to know just how to
interpret it.
There is another rationale for paying dividends, but it is not consistent with efficient
markets If managers know more about their firm than the market does and they can
time their equity issues decisions to periods when their firm is highly overvalued,
then a positive payout is optimal That is, if investors prefer constant cash flow and
managers can sell additional equity when it is overvalued, then investors will be better
off receiving a steady stream of dividends and leaving the timing of the sales to the
firm However, in efficient markets, outside investors will realize that when a firm
sells its securities, it implies that the firm is overvalued (see Myers and Majluf ( 1984),
for example), and its price (post announcement) will reflect this fact In such a case,
current equityholders are not better off, even if the managers know more about the
firm's value than the market does The attempt to raise equity will result in a reduction
in the existing equity's value The new shares will be sold at fair value, which renders
dividend policy irrelevant.
CU received a special IRS ruling so that for tax purposes, the Series A stock dividends would be
taxed in the same way as proportionate stock dividends are treated for firms having only one series of
common stock outstanding The special ruling expired in 1990.
15
Ch 7:
Payout Policy
403
A growing number of studies are presenting evidence that is not consistent with
the market rationality described above Their evidence is consistent with the notion
that managers can time the market le g , Baker and Wurgler ( 2000)l; and that the
market underreacts to some financial policy decisions, such as seasoned equity issues
lLoughran and Ritter (1995)l, Initial Public Offerings lRitter (1991) and Michaely
and Shaw (1994)l, and repurchases lIkenberry, Lakonishok and Vermaelen (1995)l.
We know that announcements of seasoned equity issues are associated with a price
decline le g , Masulis and Korwar ( 1986)l, and share repurchases announcements are
associated with price increases le g , Vermaelen (1981)l However, these studies go
further by showing that a significant price movement in the same direction continues
several years after the event.
Moreover, the post-dividend announcement drift lCharest ( 1978), Michaely, Thaler
and Womack (1995), Benartzi, Michaely and Thaler (1997)l may be a result of investor
behavior that is less than fully rational This drift can be explained to some extent by
the fact that dividend changes indicate changes in the denominator (risk profile) rather
than in the numerator (cash flows), and thus are harder to detect Grullon, Michaely and
Swaminathan (2002) find that the long-term drift is negatively related to future changes
in risk The greater the decline in risk, the larger the drift Thus, in the long run, prices
increase with a decline in risk This price behavior indicates a securities market in
which investors only gradually learn the full implications of a dividend change for
a firm's future profitability and systematic risk Hence, we could argue that paying
dividends is the optimal policy so that investors do not have to sell their stock when
it is below its (true?) market value.
The literature on dividend policy is plentiful Due to a lack of space, we cannot cover
the many contributions in detail However, one approach that has received considerable
attention in the economics literature, but not in the finance literature, was developed by
King (1977), Auerbach (1979) and Bradford ( 1981) The assumption in this framework
is that the prohibition on repurchasing shares is binding, and paying dividends is the
only way firms can distribute cash to investors The market value of corporate assets
is therefore equal to the present value of the after-tax dividends firms are expected
to pay Because dividend taxes are capitalized into share values, firms are indifferent
on the margin between policies of retaining earnings or paying dividends Thus, the
model supports the idea that firms pay out a significant portion of corporate earnings
as dividends However, this theory fails to explain the market reaction to dividend
announcements that was the starting point of many of the other theories This theory
has also not received much attention in the finance literature because of its assumption
that dividends are the only way the firm can pay out money to shareholders 16
This assumption is appropriate in some countries, such as the UK, where repurchases
have historically been illegal It is less appropriate for the USA Nonetheless, the
16 Some models have been criticized on the grounds that they implicitly assume that dividends cannot
be financed by equity or debt issues See Hasbrouck and Friend (1984) and Sarig (1984).
404
E Allen and R Michaely
use of open-market share repurchases in the USA was not common until 1983,
perhaps because of some legal restrictions For example, the risk of violating the
antimanipulative provisions of the Securities Exchange Act (SEA) of 1934 deterred
most corporations from repurchasing shares After the SEC adopted a safe-harbor rule
(Rule 10 b-18) in 1982 that guaranteed that, under certain conditions, the SEC would
not file manipulation charges against companies that repurchased shares on the open
market, repurchase activity experienced an upward structural shift.
9 Repurchases
Today, repurchases represent a significant portion of total USA corporate payouts
(Figure 1) In the last several years, the dollar amount of repurchases has been virtually
equal to that of cash dividends Not only has the amount of repurchases increased, but
also the number of firms that repurchase has increased dramatically.
The phenomenon of the decline in the number of firms that pay dividends lFama
and French (2001), Grullon and Michaely (2002)l might be directly related to the trend
we see in repurchases These trends represent a significant departure from historical
patterns in repurchase and dividend policies of corporations.
9.1 The mechanics and some stylizedfacts
Firms repurchase their shares through three main vehicles: ( 1) open-market share
repurchase, ( 2) fixed-price tender offer, and (3) Dutch auction Repurchased shares
can either be retired or be counted as part of the firm's treasury stock In any case,
those shares lose their voting rights and rights to cash flows.
In an open-market share repurchase, the firm buys back some of its shares in the
open market Historically, regulatory bodies in many countries frowned on this practice,
since it might make it possible for corporations to manipulate the price of their shares.
Indeed, there are still many countries where share repurchases are not allowed and
many other countries, such as Japan and Germany, that have only recently relaxed the
restrictions on repurchases.
In the USA, share repurchase activity is governed by the antimanipulative provisions
of the Securities Exchange Act of 1934 These provisions exposed repurchasing firms
(and anyone else involved in the repurchase activity, such as investment banks) to the
possibility of triggering an SEC investigation and being charged with illegal market
manipulation This risk seemed to deter firms from purchasing their shares Conscious
of this problem, the SEC started to design guidelines for corporations on how to carry
out share repurchase programs without raising suspicions of manipulative behavior As
part of the deregulation wave of the early 1980s, the SEC approved a legislation to
regulate open market share repurchases In 1982, the SEC adopted Rule 1Ob 18, which
provides a safe-harbor for repurchasing firms against the anti-manipulative provisions
Ch 7:
Payout Policy
405
of the Securities Exchange Act of 1934 17 Specifically, Rule l Ob-18 was adopted
in order to establish guidelines for repurchasing shares on the open market without
violating Sections 9 (a) (2) or 10(b) of the SEA of 1934
In general, Rule lOb-18
requires that firms repurchasing shares on the open market should publicly announce
the repurchase program, only use one broker or dealer on any single day, avoid trading
on an up tick or during the last half-hour before the closing of the market, and limit
the daily volume of purchases to a specified amount.
In a fixed-price tender offer, the corporation, through an investment bank, offers
to purchase a portion of its share at a prespecified price The tender offer includes
the number of shares sought and the duration of the offers However, the firm usually
reserves the right to increase the number of shares repurchased if the tender offer is
oversubscribed, and/or to buy shares from the tendering shareholders on a pro-rata
basis If the offer is not fully subscribed, the company has the right to either buy the
shares tendered or to cancel the offer altogether.
In a Dutch auction, the firm specifies the number of shares to be purchased and
the price range for the repurchase Each interested shareholder submits a proposal
containing a price and the number of shares to be tendered The firm aggregates all
the offers and finds the minimum price at which it can buy the prespecified number of
shares This price is paid to all tendering shareholders, even if they submitted a lower
price.
Table 12 shows that open-market repurchases are by far the most popular method of
repurchase For example, in 1998 open market repurchases accounted for over 95 % of
the dollar value of shares repurchased The relative importance of Dutch auctions and
tender offers, was significantly higher in the 1980s The introduction of Rule 10 b-18
and the consequent rise in the popularity of open-market share repurchases have made
the other methods much less important Therefore, in this section we concentrate on
open market share repurchases 19
In practice, fixed-price tender offers and Dutch auctions are likely to be used when
a corporation wishes to tender a large amount of its outstanding shares in a short
period of time, typically around 15 % lsee for example Vermaelen ( 1981), Comment
17 47 Fed Reg 53333 (November 26, 1982).
18 Section 9 (a) (2) establishes that it will be illegal "
to effect, alone or with one or more other
persons, a series of transactions in any security registered on a national securities exchange creating
actual or apparent active trading in such security, or raising or depressing the price of such security, for
the purpose of inducing the purchase or sale of such securities by others" Section 10(b) establishes that it
will be unlawful " to use or employ, in connection with the purchase or sale of any security registered
on a national securities exchange or any security not so registered, any manipulative or deceptive or
contrivance in contravention of such rules and regulations as the Commission may prescribe as necessary
or appropriate in the public interest or for the protection of investors".
19 Another type of share repurchase is a targeted stock repurchase, in which the firm offers to buy
stocks from a subset of shareholders For example, a "greenmail agreement" is a type of targeted stock
repurchase from (usually) one large shareholder Greenmail is typically used in conjunction with takeover
threats and is used to a much lesser extent than those described above.
406
E Allen and R Michaely
Table 12
The use of Dutch auctions, tender offers and open-market share repurchases through timea
Dutch auctions
year
Cases
Tender offers
$(mln)
year
Cases
Open market
$(mln)
year
Cases
1980
86
1,429
$(mln)
l On
Q
1 QA
I
1981
1981
44
1,329
1981
95
3,013
1982
1982
40
1,164
1982
129
3,112
1983
1983
40
1,352
1983
53
2,278
9
1984
67
10,517
1984
236
14,910
1985
1984
6
1,12 3
1985
36
13,352
1985
159
22,786
1986
11
2,33 2
1986
20
5,492
1986
219
28,417
1987
9
1,50 2
1987
42
4,764
1987
132
34,787
1988
21
7,69 5
1988
32
3,826
1988
276
33,150
1989
22
5,04 4
1989
49
1,939
1989
499
62,873
1990
10
1,93 3
1990
41
3,463
1990
778
39,733
1991
4
73 9
1991
51
4,715
1991
282
16,139
1992
7
1,63 8
1992
37
1,488
1992
447
32,635
1993
5
1,29 1
1993
51
1,094
1993
461
35,000
1994
10
92 5
1994
52
2,796
1994
824
71,036
1995
8
96 9
1995
40
542
1995
851
81,591
1996
22
2,77 4
1996
37
2,562
1996
1111
157,917
1997
30
5,44 2
1997
35
2,552
1997
967
163,688
1998
20
2,64 0
1998
13
4,364
1998
1537
215,012
1999
19
3,81 7
1999
21
1,790
1999
1212
137,015
a Source: Grullon and Ikenberry (2000), "What do we know about stock repurchase?"
and Jarrell (1991) and Bagwell (1992)l The duration of such programs is usually about
one month Open-market repurchases are often used to repurchase smaller portions of
outstanding shares, with firms repurchasing an average of 6 % of the shares lIkenberry,
Lakonishok and Vermaelen (1995), Grullon and Michaely (2002)l The duration of
open-market repurchases is much longer Stephens and Weisbach ( 1998) report that
firms complete their open-market repurchase program in about three years.
The average announcement price effect of an open-market share repurchase program
is around 3 % and the market reaction is positively related to the portion of shares
outstanding sought lIkenberry, Lakonishok and Vermaelen ( 1995), Grullon and
Michaely (2002)l Vermaelen (1981) and Ikenberry, Lakonishok and Vermaelen ( 1995)
report a decrease in stock price that is similar in magnitude in the month prior to
the announcement Comment and Jarrell (1991) report an abnormal price reaction of
around 12 % for fixed-price offers and around 8 % for Dutch auction repurchases.
Ch 7: Payout Policy
407
Using more than 1200 open market repurchases announced between 1980 and 1990,
Ikenberry, Lakonishok and Vermaelen (1995) investigated the long-term performance
of repurchasing stocks in the four-year repurchase period They found that repurchasing
firms' stock outperformed the market by an average of about 12% over the four-year
period They were particularly interested to find that most of the drift was concentrated
in "value" stocks (high book-to-market stocks) Those stocks exhibited an abnormal
return of 45 % in the four years following the repurchase announcementl
9.2 Theories of repurchases
The positive market reaction to repurchase announcements, and the fact that just like
dividends, the firms pay out cash, makes it easier to see why many of the dividend
theories apply to repurchases as well For example, we can seamlessly apply the Miller
and Rock (1985) or the Bhattacharya (1979) signaling models to repurchases At the
cost of shaving investments firms pay out cash to signal quality (Miller and Rock)
or the need for external costly financing (Bhattacharya) The free-cash-flow models
can also work as easily with repurchases as with dividends Models that are based on
relative taxation lsuch as John and Williams (1985) or Allen, Bernardo and Welch
( 2000)l or those studies that posit that dividends are a better signaling device do not
assume (or imply) that repurchases and dividends are perfect substitutes.
Before turning our attention to the substitutability of dividends and repurchases, we
review some of the work that explains why firms repurchase their shares in isolation.
Vermaelen (1984) used a standard signaling model in which managers were more
informed than outside investors about future profitability He showed that repurchasing
shares could be used as a credible signal to convey this information It is costly for
bad firms to mimic because managers hold a portion of the firm and do not tender.
Thus, if the firm buys overpriced shares and managers do not participate, the value
of their fractional share decreases Vermaelen's study also explains why the market
reaction increases with the portion of shares sought as it increases the credibility of
the signal.
Another oft-mentioned reason for buybacks relates to takeover battles By buying
back stocks from investors who value them the least, the firm makes any potential
takeover more expensive by increasing the price the acquirer will have to pay to gain
control lBagwell (1991), Stulz (1988)l The larger the fractional ownership controlled
by the management, the higher the likely premium in case of a takeover This motive
might play a role in fixed-price tender offers and Dutch auctions, in which firms
repurchase a large fraction of shares over a short period Although important in their
own right, these types of repurchase represent a very small fraction (see Table 12)
relative to open market repurchases They do not appear to be a major factor from an
overall payout policy perspective.
Repurchases can also reduce the free-cash-flow problem and mitigate conflicts of
interest between outside shareholders and management If a firm has too much cash
(beyond what it can invest in positive NPV projects), then repurchasing its shares is a
408
F Allen and R Michaely
fast and tax-effective way to give the cash back to its shareholders Moreover, buying
back shares (and assuming management has some equity, either in stocks or through
stock options) increases the relative ownership of management and decreases potential
conflicts of interest by better aligning management interests with outside shareholders'
interests las in Jensen and Meckling (1976)l.
9.3 Repurchases compared to dividends
Since dividend distributions are associated with a heavier tax burden, why not signal
or resolve agency problems only through repurchases? One answer is institutional
constraints As we noted earlier, in many countries repurchases were prohibited.
In the USA, they were limited because of regulations that subjected the firm to
manipulation charges Nevertheless, open-market repurchases were done prior to 1983,
before the introduction of Rule 10b-18 (though on a much smaller scale), and dividends
continue to be a major vehicle to distribute cash even now, nearly 20 years after the
implementation of Rule 10b-18 Some researchers have argued that if firms were to
start repurchasing shares on a regular basis, they would be challenged by the IRS.
This is another institutional constraint, but to the best of our knowledge this has not
happened yet We are not aware of any case in which the IRS has taxed a repurchase as
ordinary income on the grounds that it is a dividend in disguise, despite the fact that
a significant number of firms repurchase on a regular basis Therefore, institutional
constraints cannot be the entire story.
Several researchers have attempted to explain this puzzle from a theoretical
perspective Ofer and Thakor (1987) presented a model in which firms could signal
their value through two mechanisms, paying dividends or repurchasing their shares.
There are two types of cost associated with these signals First, by paying out cash,
firms expose themselves to the possibility of having to resort to outside financing,
which is more expensive than generating internal capital Whether a firm pays
dividends or repurchases its shares, it will be subject to this cost because these actions
deplete its internal capital The second cost, which is unique to repurchases, is that
relative to dividends, repurchases reduce managers' risk If a firm pays dividends,
which are prorated, the manager has a portion of his wealth in cash In the case of
repurchases, since she typically does not tender her shares, her portfolio is riskier.
Thus, the signaling costs through repurchases are higher It immediately follows that
if future prospects of the firms are much higher than perceived by the market, then
the managers will use repurchases If the discrepancy is not that severe, managers will
use dividends In other words, repurchases are a stronger signal.
Barclay and Smith (1988) and Brennan and Thakor (1990) provided a different
explanation as to why so many firms rely so heavily on dividends rather than
repurchases The crux of their arguments is that a portion of the firm's cost of capital
is a function of the adverse selection costs lsee Amihud and Mendelson ( 1986)
and Easley, Hvidkjaer and O'Hara (2002)l When a firm announces a repurchase
program, the cost to the uninformed investors of adverse selection increases When
Ch 7:
Payout Policy
409
some shareholders are better informed than others about the prospects of the firm, they
will be able to take advantage of this information They will bid for stock when it is
worth more than the tender price, but will not bid when it is worth less Uninformed
buyers will receive only a portion of their order when the stock is undervalued, but
will receive the entire amount when it is overvalued This adverse selection means
that they are at a disadvantage in a share repurchase When money is paid out in the
form of dividends, the informed and the uninformed receive a pro rata amount, so
there is no adverse selection As a result, uninformed shareholders prefer dividends
to repurchases Further, this preference will persist even if dividends are taxed more
heavily than repurchases, provided the tax disadvantage is not too large On the other
hand, the informed will prefer repurchases because this allows them to profit at the
expense of the uninformed.
Brennan and Thakor (1990) argue that the method of disbursement chosen by firms
will be determined by a majority vote of the shareholders If the uninformed have more
votes than the informed, firms will use dividends, but if the informed predominate,
firms will choose repurchases When there is a fixed cost of obtaining information,
the number of informed depends on the distribution of shareholdings and the amount
paid out For a given payout, investors with large holdings will have an incentive
to become informed When a small amount is paid out, only the investors with the
largest holdings will become informed; most shareholders will remain uninformed and
will prefer dividends When a larger amount is paid out, more shareholders become
informed, so the firm may choose repurchases.
We note that this model has exactly the opposite prediction to Allen, Bernardo and
Welch (2000) on the relation between large (and presumably informed) shareholders
and payout policy In this model, larger shareholders favor repurchases In Allen et al ,
large shareholders prefer dividends It is still an open question as to which one of these
predictions holds empirically.
The Brennan and Thakor model is an intriguing explanation of the preference that
firms appear to have for dividends It answers the question of why firms prefer to
use dividends even though dividends are taxed more heavily Unlike the John and
Williams' theory, the Brennan and Thankor model supports the idea that dividends
are smoothed.
However, their model is not above criticism First, the range of tax rates for
which dividends are preferred to repurchases because of adverse selection is usually
small To explain the predominance of dividends, we must use another argument that
relies on shareholders being homogeneous For tax rates above the level at which
adverse selection can explain the preference for dividends, everybody will tender in a
repurchase, so it will be pro rata But this universal tendering clearly does not occur.
Second, if superior information is the motive for repurchases, it is surprising that
management almost never tenders its shares Presumably, they are the ones with the
best information Another criticism is that if adverse selection were a serious problem,
firms could gather the relevant information and publicly announce it Nevertheless,
410
F Allen and R Michaely
Brennan and Thakor's theory sheds new light on the choice between dividends and
repurchases.
Chowdhry and Nanda ( 1994) and Lucas and McDonald ( 1998) also considered
models in which there is a tax disadvantage to dividends and an adverse selection cost
to repurchases In their models, managers are better informed than are shareholders.
Their models show how payout policy depends on whether managers think the
firm is over or undervalued relative to the current market valuation Both models
provide interesting insights into the advantages and disadvantages of dividends and
repurchases However, the stability and smoothing of dividends is difficult to explain in
this framework unless firms remain undervalued or overvalued relative to their market
value through time.
9.4 Empirical evidence
The market usually reacts positively to an announcement of any type of share
repurchase The extent of the reaction is positively related to the size of the repurchase
program and negatively related to the market value of the firm Despite the positive
reaction, many studies have found that the market does not comprehend the full
extent of the information contained in the announcement, given the long-term postannouncement drift The drift is particularly pronounced in high book-to-market stocks
lfor open-market share repurchases, see Ikenberry, Lakonishok and Vermaelen (1995)l.
Vermaelen (1981), Comment and Jarrell (1991), Ikenberry, Lakonishok and Vermaelen
(1995), and others document a negative abnormal return in the months leading to the
(open market) repurchase announcement, a finding that suggests that firms time the
repurchase announcement to when the stock is more undervalued.
A subtler issue concerns the number of shares that have actually been repurchased
and the duration of the program A firm is under no obligation to repurchase all of the
shares it seeks The announcement merely serves to inform investors of its intentions.
If there is a significant discrepancy between the announced and the actual number of
shares repurchased, this discrepancy can affect the long-term reaction in the years after
the announcement Just as important, when we wish to examine the relation between
repurchases and other types of payout such as dividends, or to relate actual repurchases
to performance, we must measure the actual repurchases as accurately as possible.
9.4 1 How to measure share repurchase activity?
Using 450 open-market repurchase programs announced between 1981 and 1990,
Stephens and Weisbach (1998) suggest several measures of repurchases.
(1) The change in number of shares outstanding as reported on the CRSP or Compustat
databases.
A potential problem with this measure is that if a firm repurchases shares and
simultaneously distributes shares (either to the public or to employees), this measure
will understate the actual amount of repurchase.
Ch 7:
Payout Policy
411
( 2) The net dollar spent on repurchases as reported in the firm's cash flow statement.
If we want to analyze the dollar amount spent on repurchases, this measure is probably
the best one to use If we wish to compute the number of shares repurchased, we must
convert the dollar number that is reported in the cash flow statement to number of
stocks repurchased However, doing so creates a difficulty, since we do not know the
purchase price We can use the average trading price over the period as a proxy for
the purchase price Another possible shortcoming of this measure is that it includes
purchases of not only common stocks, but also other type of stocks such as preferred
stocks However, repurchases of securities other than common stocks represent only a
very small portion of firms' repurchase activity.
(3) The change in Treasury stock (also reported on Compustat).
However, this measure can be problematic, since firms often retire the shares they
repurchase Thus, while the number of shares outstanding decreases, the number of
Treasury shares does not change In addition, if a firm repurchases shares and at the
same time distributes shares, say in lieu of stock options, there is no change in Treasury
stock, despite the repurchase activity This factor may represent a significant problem,
given the recent popularity of stock options as a method of compensation.
For example, imagine a firm that repurchases 1000 shares, say for $10000, and
then a few months later turns around and give these shares to its CEO as part
of her compensation The firm is involved in two distinct actions The first is a
financing action (repurchasing shares), and the second is an investment decision
(paying the manager) If we try to analyze the impact of a financing decision, holding
all else constant, especially holding investment constant, this measure of repurchase
is inadequate.
The problem is even more severe if we try to compare repurchases and dividend
decisions Say, our firm pays a total dividend of $10000, instead of repurchasing its
shares At the same time, it also issues shares and gives them to the manager In the
first case (when the firm repurchases its shares in the open market and the researcher is
using Treasury shares to measure repurchases), we would record no repurchase activity.
But in the second case (pay a dividend and issue shares), we would record a $10000
dividend But in reality, assuming away taxes, both routes are exactly identical Our
firm pays $ 10000 to shareholders and gives $ 10000 worth of stock to the manager.
In summary, measuring repurchases through the change in Treasury stock is likely to
yield the most biased measure of repurchases It can bundle investment and financing
decisions (as discussed above), it combines other overlapping distributions, and it does
not account for the fact that many firms retire the stocks they repurchase rather than
putting them into Treasury stock Stephens and Weisbach (1998) find that this measure
is substantially different from the other measures they use They show that the first
two measures yield similar results in the measurement of share repurchases, while the
Treasury-stock method yields estimates that are lower than the other two methods by
about 60 %.
Which method should we use? We recommend using the cash flow spent on
repurchases, and trying to account for any changes in the shares outstanding This
412
E Allen and R Michaely
measure is likely to yield the least biased estimate of the actual dollar amount spent
on repurchases.
Given these measures of actual repurchases, we can address the issue of how long
it takes firms to complete their announced stock repurchase program Stephens and
Weisbach (1998) reported that approximately 82 % of the programs were completed
within three years More than half of the firms completed their announced repurchase
program, but one tenth of the firms repurchased less than 5% of their announced
intentions The authors also showed that the initial market reaction to share repurchases
was positively related to the actual share repurchase activity in the two years after
the announcement Firms that repurchased more, experienced a larger positive price
effect at the announcement However, the announcement effect was not related to the
announced quantity of share repurchase.
Stephens and Weisbach ( 1998) also showed that the actual amount of repurchase
in a given quarter was related to the firm's cash flow level Using a Tobit model,
they showed that the decision to repurchase was positively related to both the level
of expected cash and unexpected cash They also showed that the actual repurchase
activity was negatively related to the equity return in the previous quarter: the more
negative the return was in quarter t 1, the more likely the firm was to engage in
repurchase activity in quarter t.
9.4 2 Empirical tests of repurchase theories
So repurchases are positively greeted by the market, they are preceded by bad
performance, and some (mainly value stocks) are followed by positive abnormal
price performance All of these attributes are consistent with both the asymmetric
information/signaling and the free cash flow theories as the main motive behind the
decision to repurchase But, as with dividends, there are two possibilities The positive
price impact of the announcement can be because repurchases are good news (i e , they
lead to better investment decisions because management has less cash to squander), or
repurchases can convey good news (i e , they do not change investment decisions, but
they merely convey that the firm's future growth in cash flows are under-valued) The
negative price performance in the months before the announcement and the positive
price performance in the years after also support both explanations The stock price
might have increased either because the market did not comprehend the full extent of
the undervaluation, or because it did not incorporate the extent of the better investment
decisions by management after the repurchase.
Thus, to determine the dominant force behind the decision to repurchase, we
must look elsewhere We begin with Vermaelen (1981) Using a number of fixedprice tender offers over the period 1962-1977, Vermaelen documented a significant
increase in earnings per share in the years following fixed-price repurchases Using
122 observations from a similar period, Dann, Masulis and Mayers ( 1991) confirmed
Vermaelen's findings They also showed that the initial market reaction was positively
related to subsequent increases in earnings Although a decline in cash flows (or
Ch 7:
Payout Policy
413
earnings) in the years after fixed-price tender offers will lead to a rejection of the
information/signaling hypothesis, these studies found that an increase in earnings was
consistent with the information/signaling hypothesis.
However, in a detailed investigation of 242 fixed-price tender offers, Nohel and
Tarhan (1998) showed that the entire improvement in earnings documented in previous
studies could be attributed to firms with high book-to-market That is, to low-growth
value firms Furthermore, they showed that firms involved in tender offers did not
increase their capital expenditure, and in fact that the improvement in operating
performance of the high book-to-market firms was positively related to asset sales.
This finding was inconsistent with the signaling model They interpreted their evidence
as supporting the notion that fixed-price tender offer, and the market reaction to them,
is motivated by free cash-flow considerations rather then signaling.
The earnings pattern after open-market share repurchases shows an even more
consistent lack of improvement than those after fixed-price tender offers Grullon
and Michaely ( 2003) examined a comprehensive sample of 2735 open-market share
repurchases in the period 1980-2000 They reported a decline in the level of
profitability (measured by ROA) in the three years after the year in which the
repurchase was announced 20 They also reported a decline in capital expenditures and
cash reserves for those firms (Using a different sample, Jagannathan and Stephens
(2001) reach similar conclusions) Overall, it seems that earnings performance
subsequent to open-market repurchase programs and earnings performance after large
changes in dividends have a very similar pattern.
The risk profile of firms changes in conjunction with open-market share repurchases -just as it changes after dividend increases Grullon and Michaely ( 2003) found
that beta declined in the year after the announcement The cost of capital in the three
years after open-market repurchases declined significantly from an average of 16 3%
before the repurchase to 13 7 % after 21
The evidence of declining earnings, a reduction in capital expenditures and cash
reserves, and a decline in risk is not consistent with the traditional signaling stories It
is consistent with the notion that when investment opportunities shrink and there is less
need for capital expenditures in the future, firms increase their payout to shareholders,
either through dividends or through open-market share repurchases Thus, when a
firm is in a different stage of its life cycle, its investment opportunities change, and
consequently its risk profile and need for cash changes as well This change in turn
Using a sample of 185 open-market share repurchases over the period 1978-1986, Bartov (1991)
reported mixed results on the relation between earnings changes and repurchases In the year after the
open-market repurchase, those firms' earnings were significantly worse then the control sample In the
year after that, they were significantly better These mixed results might be attributable to the small
sample size.
21 Other studies found a similar phenomenon with fixed-price tender offers See Dann, Masulis and
Mayers (1991), Hertzel and Jain (1991) and Nohel and Tarhan (1998) These studies showed that the
market reaction to the offer is positively related to the subsequent decline in risk.
20
414
EFAllen and R Michaely
affects it payout policy, because it increases dividends, repurchases or both (It is still
an open question what determines the form of payout a firm chooses to use )
Some of the evidence in Ikenberry, Lakonishok and Vermaelen ( 1995)22 also
supports this notion They reported that the largest price appreciation in the years
after the repurchase occurred for those firms that were most likely to benefit from
disposing of cash Those firms with high book-to-market ratio were the firms that had
less need for future capital expenditure and were more likely to encounter free cashflow problems.
This is not to say that perceived undervaluation does not play a role at least in the
timing of the repurchase programs Many of the studies cited above show that there
is a clear tendency for firms to repurchase shares after a decline in stock price, which
suggests that management repurchases shares when they think the stock is undervalued.
An extreme example is the heavy wave of share repurchases immediately after the stock
market crash of October 1987.
In addition, Ikenberry, Lakonishok and Vermaelen (2000) provided evidence that in
value stocks and small cap stocks, management bought more shares when the price
dropped and fewer shares when the price rose What is clear from their evidence is
that this undervaluation is not related to future earnings growth It may happen because
of changes in the risk profile of the firm that are not impounded in market price It
might be that for value stocks that have not performed well in the past, investors are
more reluctant to believe that these firms will turn around, cut capital expenditure,
reduce the amount of cash reserves, and reap the benefits of reductions in free cash
flows Hence, ex-post, those stocks outperform their peers when information about the
realization of these issues starts to appear in the market place.
Miller and McConnell (1995) studied adverse selection as a motive for repurchases
by examining one of the direct implications of Barclay and Smith's (1988) conjecture
and the Brennan and Thakor (1990) model These theories argued that corporations
relied on dividends rather than repurchases because of adverse selection problems.
When a firm announces a share repurchase program, the uninformed market participants, particularly the market makers, should assume that they are more likely to trade
with informed traders Hence, in response to this signal, the bid-ask spread should
widen Using daily closing quotes around 152 open market share repurchase programs,
Miller and McConnell found no evidence of an increase in bid-ask spread that they
could associate with repurchases There was no evidence that firms were deterred from
engaging in open market share repurchase programs because of the adverse effect of
such programs on market liquidity or on the firm's cost of capital Moreover, Grullon
and Ikenberry ( 2000) presented evidence that share repurchase programs enhanced
liquidity, rather than reducing it.
22 Ikenberry, Lakonishok and Vermaelen (2000) reported similar results for Canadian open-market
repurchases.
Ch 7: Payout Policy
415
The empirical evidence indicates that repurchase activity is motivated by several
factors Firms with more cash than they need for operation (excess cash) are more
likely to repurchase their shares Lower-growth firms are more likely to repurchase
shares, because their investment opportunities shrink Researchers find evidence that
both the announcement of repurchases and the actual repurchase activity is more
pronounced at times when firms experience downward price pressure There is no
evidence that adverse selection in the market place is a reason for repurchases, nor
is there any evidence that the market's underestimation of future cash flows or growth
in earnings (or cash flows) are a motive in management's decision to repurchase In
fact, the evidence shows that repurchasing firms experience a reduction in operating
performance, have excess cash, and invest less in the years after the repurchase
announcement, and that their risk is significantly lower in the post-announcement
years.
It is also clear that the market does not incorporate the entire news contained in
the repurchase announcement, be it about risk reduction, reduction in agency costs,
or some other misvaluation The market underreaction is particularly pronounced for
value stocks.
9.4 3 Some empirical evidence on dividends compared to share repurchases
Equipped with the measures of actual repurchases that we discussed above, researchers
were able to examine the issue of how dividend and repurchase policies interact It was
also possible to consider whether firms view these methods as substitutes.
Many of the theories discussed above have implications to whether repurchases and
dividends are substitutes, or if they are used for different objectives altogether, which
would indicate that there is no relation between dividends and repurchase policies.
Theories that address the issue of total payout policy, such as Miller and Rock
( 1985) or Bhattacharya (1979), and which make no distinction between dividends and
repurchases, imply that these two payout policies are perfect substitutes Other theories,
which rely on differential taxation, such as those by John and Williams (1985) and
Allen, Bernardo and Welch (2000), imply that these two payout policies are distinctly
different and that there cannot be direct substitution between the two.
The agency theories also imply substitution, but the substitution is not perfect.
On the one hand, both repurchases and dividend payments take money out of
management's hands and thereby reduce potential abuses On the other hand, dividends
act as a stronger commitment device, because management is more committed to
maintaining a stable dividend policy than a stable repurchase policy lsee Lintner
(1956)l Thus, it is possible that management might distribute temporary excess cash
through repurchases and more permanent excess cash through dividends.
There is another reason why managers may have an incentive to pay fewer dividends
and distribute more of the cash in the form of repurchases This is the growing
popularity of stock options, and especially of executive stock options Stock options
can affect the form of payment for at least two reasons First, since these options
416
E Allen and R Michaely
are typically not protected against dividends, managers (who own stock options) have
an incentive to repurchase shares with the available cash rather than pay it out in
the form of dividends Second, many market analysts center their stock valuation
on EPS numbers Since the exercise of stock options dilutes EPS numbers, both the
boards of directors and top management may decide to repurchase more shares to
prevent dilution 23 Thus, stock options can lead to the substitution of dividends for
repurchases.
We could argue that by definition, dividends and repurchases are perfect substitutes.
A firm can either pay dividends or repurchase its shares If, and only if, total payout
is held constant is this statement correct But we already know that all else is not
constant Firms can change the amount of cash kept in the firm, they can alter the
amount of cash that goes to investments, and they can change the amount of cash that
they raise from other sources, such as debt or equity.
Therefore, another way to pose the question is to ask what has happened to total
corporate payout since repurchases have become so popular Have dividends been
reduced correspondingly so that total payout remains at a constant level? Or has
total payout increased? Whether the increased popularity of repurchases increased
corporate payout can be critically important to corporations, investors, and policy
makers alike The answer to this question has significant implications concerning
corporate reinvestment rates, resource allocation, and the role of taxes in corporations'
decisions But despite its importance, only recently has the issue begun to receive
attention from financial economists.
An analogous question has been recently debated in the public finance literature.
The issue is the impact of 401 k and IRA programs on USA saving rates, where 401 k
is the equivalent of repurchase programs and the total saving rates is analogous to total
payout Has the introduction of these saving programs increased savings rates, or has
it merely caused a shift from one saving vehicle to another? (See Poterba, Venti and
Wise (1996) for an excellent review of this issue).
In both cases (saving rates and payout rates), the key impediment to determining the
impact of IR As on saving and repurchases on payouts is agents' heterogeneity Some
corporations pay cash (mostly, the mature firms) and some corporations (those firms
with growth opportunities) do not pay out cash to shareholders Those that do pay tend
to pay more in both forms Thus, one of the main challenges for such an investigation
is to control for this heterogeneity in various ways.
In Table 1 and in Figure 1 we presented the pattern of dividends, repurchases, and
total payout of USA industrial corporations through time relative to total corporate
23 We do not to argue that this reason is rational (or irrational) It seems to be the case however, that this
is a driving force behind many corporate financial decisions For example, both authors of this chapter
have heard on numerous occasions that one of the important yardsticks of mergers to be consummated is
its impact on EPS Managers are very reluctant to enter into a merger or an acquisition that dilutes EPS.
Likewise, the impact of repurchases on EPS is also often mentioned See also the discussion in Dunbar
(2001) of how British institutional investors impose dilution constraints on management.
Ch 7:
417
Payout Policy
earnings and relative to the corporations' market value The table shows that relative
to total earnings, total payout has increased through time It also shows that dividend
payout did not decrease, despite the surge in repurchases However, when we scale the
cash payout by market value (Figure 1), the opposite picture emerges Dividend yield
has been going down through the years and repurchase yield has been going up At
least through the 1990s, there is no change in the total payout yield.
However, the aggregate data may mask a qualitative difference across firms For
example, there could be some firms that never paid dividends and have recently started
to pay out cash in the form of repurchases At the same time, firms that have been
paying dividends might have continued to do so.
To address the interaction between repurchase and dividend policy, Grullon and
Michaely (2002) examined this relation at the individual-firm level as well Their test
relies on Lintner's (1956) analysis of how firms determine their dividend policy Lintner
observed that firms' dividend change decisions were a function of their targeted payout
ratio and the speed of adjustment of current dividends to the target ratio Using this
model, Grullon and Michaely calculated the expected dividend payment for a firm
based on its past dividend behavior, and determined whether actual dividend payments
were above or below the expected dividend payment That way, they were able to
observe whether a firm was deviating from its past dividend policy If the use of
repurchases increased payout and did not affect dividend policy, then there would
not be any relation between the dividend forecast error from the Lintner model and
repurchase activity Grullon and Michaely defined the dividend-forecast error as:
ERROR,i =
ADIV,i
(l,
+
2,iEARNt,i +
f33DI
,i Vt -i)
M Vt l~i
where ADIV,ti is the actual change in dividends at time t; EARN,,i is the earnings
at time t; DIV,_ l i is the dividend level at t 1; and MV,_ l i is the market value of
equity at time t 1 The coefficients 32,i and 3,i are the parameters of earnings and
lagged dividends from Lintner's ( 1956) model, respectively, that have been estimated
over the pre-forecast period, 1972-1991 By scaling by the firm market value of equity,
they were able to directly compare the forecast error to the repurchase and dividend
yields.
However, if repurchase activity reduces dividend payout, then the test should have
result in a negative correlation between the dividend forecast error (actual minus
expected) and share repurchase activity In other words, finding a negative correlation
between these two variables would indicate that share repurchases have been partially
financed with potential dividend increases.
Their empirical evidence indicates that the dividend forecast error is negatively
correlated with the share-repurchase yield The forecast error becomes more negative
(monotonically) as the share repurchase yield increases That is, as firms repurchase
more, the actual dividend is lower than the expected dividend.
They confirmed this result by a cross-sectional regression of the dividend-forecast
error on the repurchase yield (controlling for size, the return on assets, the volatility
418
E Allen and R Michaely
of return on assets, and nonoperating income) The results indicate that the repurchase
yield has a negative effect on the dividend forecast error even after controlling for firm
characteristics.
In summary, the evidence suggests that dividend-paying firms have been substituting
dividends with share repurchases, but the rate of substitution is not one (i e , they are
not perfect substitutes) This finding supports the idea that share-repurchase policy and
dividend-policy are interrelated.
But what types of firms use, and under which circumstances would managers decide
to use, repurchases and/or dividends? We do not yet have the complete picture, but
some recent research gives us some idea.
The first issue is the relation between stock-option programs and payout policy.
Incentive compensation such as stock options could affect total payout if it aligns
management incentive with those of shareholders, and therefore induces management
not to invest in value-destroying projects and pay more to shareholders Thus, incentive
compensation may increase total payout Additionally, as suggested before, managers
with stock options, which are not dividend-protected, will be motivated to shift the
form of payout from dividends to repurchases.
Using a large sample of 1100 nonfinancial firms during the period 1993-1997,
Fenn and Liang (2001) reported a negative relation between stock-option plans and
dividends, a finding that supports the notion that the use of managerial incentive
plans reduces managers' incentive to pay dividends Moreover, their cross-sectional
regression results indicated that (1) dividend payout was negatively related to the
magnitude of stock-option plans; (2) repurchase payout was positively related to
the magnitude of stock-option plans; and ( 3) total payout was negatively related to the
magnitude of stock-option plans The reduction in total payout was larger than the
increase in repurchases.
Using a sample of 324 firms that announce a change in payout policy in 1993, Jolls
(1998) found a positive relation between the repurchase decision and the magnitude
of the executive stock-option plan.
Weisbenner (2000) extended these studies He asked if the group holding the stock
options (the firm's employees or management) made a difference on payout choice.
A priori, we would expect it to do so If mainly nonexecutive employees hold stock
options, then the dividend protection is less of a factor (assuming management does
not maximize employees' wealth) The dilution factor is still important, since it affects
everyone who holds the stock, not just the employees Thus, in the case of nonexecutive
stock option plans we would expect an increase in repurchase activity but no reduction
in dividends If executives hold stock options, then we should expect both a reduction
in dividends and an increase in repurchase activity.
Weisbenner (2000) found empirical support for these hypotheses The overall size
of a firm's stock option program had a significant influence on the firm's repurchase
policy (presumably in an attempt to prevent dilution) Stock-option programs are also
related to the firm's propensity to reduce retained earnings Second, the larger the
executives' holding of stock options, the more likely the firm was to reduce dividends
Ch 7:
Payout Policy
419
and to retain more of its earnings (presumably an outcome of managers' incentive not
to pay dividends).
The studies discussed above show an important link between compensation, and
executive compensation in particular, and the form of payout As the extent of
stock option programs increase, firms tend to use more repurchases and to reduce
retained earnings When more of these stock option programs are directed towards
top management, dividends also tend to be reduced.
Jagannathan, Stephens and Weisbach ( 2000) found another important link between
firm's characteristics and payout policy As with Lintner's model, the authors
hypothesized that dividends were more of a permanent commitment than were share
repurchases Hence, dividends were more likely to be paid out of permanent earnings
and repurchases were more likely to be used as a way to distribute temporary cash
flows The empirical implication of this hypothesis is that firms that experience
higher cash flow variability tend to use repurchases while firms with lower cash flow
variability tend to use dividends.
Using a large sample of repurchase and dividend change events, Jagannathan,
Stephens and Weisbach (2000) found that firms that repurchased their shares had a
higher variability of operating income relative to firms that only increased dividends,
or to firms that increased their dividend and repurchased their shares Not surprisingly,
they found that firms that did not pay cash had the highest cash flow variability of
all Using a Logit model, they showed that higher cash flow variability and higher
nonoperating cash flow (two measures of temporary earnings) increased the likelihood
of repurchases relative to dividends As had earlier studies, they also found that
although dividends appeared to be paid out of permanent earnings, there was no
evidence of earnings improvements following dividend increases.
Lie's ( 2001) results also pointed in the same direction He found that tender offers
were more likely to occur when firms had excess cash on their balance sheet (a
temporary build-up of cash), and dividends were more likely to increase with excess
cash on the income statement (presumably a permanent increase in cash flow).
Overall, the evidence indicates that at least in cross-sectional tests, firms that use
stock options more intensely are more likely to use share repurchases The evidence
also associates firms that only repurchase with firms that are riskier (relative to those
who pay dividends and those who do both) There is also some evidence that the
increase in popularity of repurchases might be related to changes in regulation The
extent to which these variables can explain the dramatic increase in repurchases and
the more moderate increase in overall payout is still an open question.
9.5 Summary
Open-market repurchases have become a dominant form of payout Given the
economic climate and the deregulation of repurchasing shares around the world, we
believe that the phenomenon is here to stay Repurchases are likely to remain a
dominant form of payout from corporations to their shareholders As researchers, we
420
F Allen and R Michaely
do not yet have a clear grasp on how firms decide among the various forms of payouts,
and in particular, how they decide on whether to pay cash in the form of dividends or
share repurchases Nor do we know how the decision affects their retained earnings
and their investment decisions.
The empirical evidence starts to give us some directions It seems that young,
risky firms prefer to use repurchases rather then dividends, though we do not fully
understand what determines the choice We observe that many large, established firms
have substituted repurchases for dividends That is not to say that those firms have
necessarily cut the nominal dividends, but they have increased dividends at a much
lower rate than before Instead, they have been paying more money to shareholders
through repurchases We see that those firms with more volatile earnings tend to
substitute more often But again, we do not have a firm understanding of what
determines that choice Finally, we ask how repurchases and payout policy as a whole
interact with capital structure decisions (such as debt and equity issuance) We believe
that these are very important questions and a promising field for further research.
10 Concluding remarks
There are a number of important empirical regularities concerning firms' payout policy.
The first is that the mid-1980 S represented a watershed Earlier, dividends constituted
the vast majority of corporate payouts They grew at an average of about 15 % per year.
Dividend yields over the long run remained fairly constant There were repurchases,
but they represented only a small fraction of payouts.
Since the mid 1980 s, repurchases have become increasingly important Dividends
have continued to increase in absolute terms, but at an average rate of 6 % rather than
15 % a year Instead of increasing dividends, companies have been much more willing
to increase the absolute payout by increasing repurchases Repurchases have grown
steadily and are now about the same level of magnitude as dividends The result of
these changes is that in the last decade or so, dividend yield has fallen significantly
from 3 % to 1 5 %, but the yield resulting from the combination of dividends and
repurchases has remained fairly constant at 3 %.
At the level of the individual firm there are a number of interesting regularities.
Although dividends have decreased in relative importance and firms are much more
willing to switch to repurchases, dividends are still important in absolute terms Firms
seem reluctant to cut dividends However, firms that have never paid dividends do not
seem to regard them as a necessity Over the years, firms that initiate payments do so
increasingly through repurchases In the last five years, about 75 % of initiating firms
have used this method of payout.
Another important aspect of the comparison between dividends and repurchases
is that both have similar effects in terms of the sign of the impact Initiation of
dividends, dividend increases, or repurchases are all taken as good news by the market.
Ch 7:
Payout Policy
421
The difference is that repurchases are larger in size relative to dividend increases or
initiations, and their impact on prices is more pronounced.
Although these empirical regularities seem clear and provide a guide for how
managers should behave, our understanding of why firms behave in this way is, to say
the least, limited This is the case despite the enormous effort that has been invested
in the topic of payout policy over the years It is possible to tell a story, but it is by
no means clear that it is anything more than a story.
If we go back over a century or more, there seem to be obvious advantages
to paying dividends Information was sparse and any firm that could consistently
pay out dividends was signaling that it had long-term earning potential Firms that
constantly repurchased and intervened in the market for their shares may well have
been suspected of manipulating the stock price Moreover, for individuals to sell shares
was an expensive business in terms of direct transaction costs Extensive insider trading
and other similar abuses meant that, in terms of adverse selection, there was also a
significant short-term cost from selling This environment established a convention that
paying dividends was good and cutting dividends was bad.
The change in the laws concerning repurchases and stock-price manipulation in
1982 meant that repurchases could be used without risk and made them an acceptable
alternative However, since cutting dividends is perceived as a bad signal, at least in
the short run, firms are not willing to replace dividends with repurchases even though
repurchases have tax advantages However, as payout is increased, repurchases can be
increasingly used.
The other piece of the payout puzzle is that total payout yield in terms of dividends
and repurchases has remained fairly constant at least for the last ten years One possible
explanation for this is a signaling story The market treats increases in dividends and
repurchases as good news In theory, this reaction could be because increases are
interpreted as signals of future operating performance However, there is evidence
that increases in payout are not followed by improved operating performance, thus
rejecting this explanation An alternative interpretation is that the market is relieved
that managers will no longer acquire cash that can be squandered, and this is why an
increase in payout leads to a higher share price.
Of course, all of this argument ignores many important factors, but it is an example
of one explanation for the patterns that are observed in the data Much work remains
to be done.
So far, our discussion here has focused on dividends and repurchases But there is
a third component of payout that has been largely ignored in the literature, and that is
the cash payments for securities acquired in M&A transactions The precise amount
paid out in this way is difficult to measure exactly However, the data we have gathered
that does allow us to establish a lower bound suggests that over the last decade, such
payments have been around $240b per year, or over 50 % of aggregate payout if we
also include dividends and repurchases Measuring and understanding this component
of payout policy is an important task for future research.
422
E Allen and R Michaely
At this stage, we cannot recommend an optimal payout policy However, we can
make several general (and, admittedly, somewhat speculative) suggestions:
(1) Following the example of the last decade, repurchases should be used much
more frequently than they have been Investment and repurchase policies should
be coordinated to avoid the transaction costs of financing When there are
positive NPV investments, repurchases should be avoided In years where
NPV investment opportunities are low, unneeded cash should be paid out by
repurchasing shares.
(2) To the greatest extent possible, firms that have a high degree of information
asymmetry and large growth opportunities should avoid paying dividends The
significant costs associated with raising equity capital for these firms makes
payment of dividends even more costly Stated differently, in periods when a firm
faces many good investment opportunities, a dividend reduction might not be such
a bad idea.
( 3) Given the restrictive dividend-related covenants and the fact that firms interact
with bondholders more than once, the use of dividends to extract wealth from
debtholders should be avoided Most times, it does not work Even when it does,
the long-run result can be detrimental to equityholders (There is no evidence that
management follow this strategy in practice )
(4) We cannot think of a good reason why most USA firms pay dividends on a
quarterly basis instead of on an annual basis Longer intervals between payments
would allow investors that are interested in long-term capital gains to sell the
stock before the ex-day, avoid paying tax on the dividend, and maintain the longterm tax status of the stock Such a schedule would also allow corporations who
might be interested in dividend income to minimize transaction costs and deviation
from optimal asset allocation while capturing the dividend Finally, it would save
the dividend-paying corporation administrative and mailing costs associated with
dividend payments.
(5) Avoid costly "signals" Hopefully, the firm is going to stay alive for a long time.
Managers can find cheaper and more persuasive ways to credibly convey the
company's true worth to the market.
(6) The difference in taxes between dividends and capital gains makes high-yield
stocks less attractive to individual investors in high tax brackets Such investors
should try to hold an otherwise identical portfolio with low-yield stocks.
Other people might disagree with these suggestions However, until our understanding of the subject is improved, they represent a logical way for managers and investors
to proceed Much more empirical and theoretical research on the subject of payout is
required before a consensus can be reached.
References
Ackert, L , and B Smith (1993), "Stock price volatility, ordinary dividends, and other cash flows to
shareholders", Journal of Finance 48(4):1147-1160.
Ch 7:
Payout Policy
423
Aharony, J , and I Swary (1980), "Quarterly dividend and earnings announcements and stockholders'
returns: an empirical analysis", Journal of Finance 35 (1):1-12.
Allen, E, and R Michaely (1995), "Dividend policy", in: R Jarrow, V Maksimovic and WT Ziemba,
eds , Handbook in Operations Research and Management Science, Vol 9 (Elsevier, Amsterdam).
Allen, E, A Bernardo and I Welch (2000), "A theory of dividends based on tax clientele", Journal of
Finance 55 (6):2499-2536.
Ambarish, R , K John and J Williams (1987), "Efficient signaling with dividends and investments",
Journal of Finance 42(2):321-343.
Amihud, Y , and H Mendelson (1986), "Asset pricing and bid-ask spread", Journal of Financial Economics
17:223-250.
Amihud, Y , and M Murgia (1997), "Dividends, taxes, and signaling: evidence from Germany", Journal
of Finance 52(1):397-408.
Asquith, P , and D W Mullins Jr (1983), "The impact of initiating dividend payments on shareholders'
wealth", Journal of Business 56(1):77-96.
Auerbach, A J (1979), "Wealth maximization and the cost of capital", Quarterly Journal of Economics
v93 (3):433-446.
Bagwell, L (1992), "Dutch auction repurchases: an analysis of shareholder heterogeneity", Journal of
Finance 47(1):71-105.
Bagwell, L S (1991), "Shareholder heterogeneity: evidence and implications", American Economic
Review 81(2):218-221.
Bagwell, L S , and J Shoven (1989), "Cash distributions to shareholders", Journal of Economic
Perspectives 3(3):129-140.
Baker, M , and J Wurgler (2000), "The equity share in new issues and aggregate stock returns", The
Journal of Finance 55:2219-2257.
Bali, R , and G Hite (1998), "Ex-dividend day stock price behavior: discreteness or tax-induced
clientele?", Journal of Financial Economics 47:127-159.
Barclay, M (1987), "Dividends, taxes, and common stock prices: the ex-dividend day behavior of
common stock prices before the income tax", Journal of Financial Economics 14:31-44.
Barclay, M J , and C W Smith Jr (1988), "Corporate payout policy: cash dividends versus open-market
repurchases", Journal of Financial Economics 22(1):61-82.
Bartov, E (1991), "Open-market stock repurchase as signals for earnings and risk changes", Journal of
Accounting and Economics 14:275-294.
Benartzi, S , R Michaely and R Thaler (1997), "Do changes in dividends signal the future or the past?",
Journal of Finance 52(3): 1007-1043.
Bernhardt, D , J E Robertson and R Farrow (1994), "Testing dividend signaling models", Working Paper
(Queen's University, Ontario, Canada).
Bernheim, D (1991), "Tax policy and the dividend puzzle", Rand Journal of Economics 22:455-476.
Bernheim, D , and A Wantz (1995), "A tax-based test of the dividend signaling hypothesis", American
Economic Review 85(3):532-551.
Bhattacharya, S (1979), "Imperfect information, dividend policy, and 'the bird in the hand' fallacy",
Bell Journal of Economics 10(1):259-270.
Binay, M (2001), "Do dividend clienteles exist? Institutional investor reaction to dividend events",
Working Paper (University of Texas, Austin).
Black, F (1976), "The dividend puzzle", Journal of Portfolio Management 2:5-8.
Black, E, and M Scholes (1974), "The effects of dividend yield and dividend policy on common stock
prices and returns", Journal of Financial Economics 1:1-22.
Blume, M E (1980), "Stock return and dividend yield: some more evidence", Review of Economics
and Statistics 62:567-577.
Blume, M E , J Crockett and I Friend (1974), "Stock ownership in the United States: characteristics
and trends", Survey of Current Business, pp 16-40.
424
E Allen and R Michaely
Boyd, J , and R Jagannathan (1994), "Ex-dividend price behavior of common stocks: fitting some pieces
of the puzzle", Review of Financial Studies 7(4):711-741.
Bradford, D E (1981), "The incidence and allocation effects of a tax on corporate distributions", Journal
of Public Economics 15:1-22.
Brav, A , and J B Heaton (1998), "Did ERISA's prudent man rule change the pricing of dividend omitting
firms?", Working Paper (Duke University, NC).
Brennan, M J (1970), "Taxes, market valuation and financial policy", National Tax Journal 23:417-429.
Brennan, M J , and A V Thakor (1990), "Shareholder preferences and dividend policy", Journal of
Finance 45 (4):993-1019.
Brickley, J (1983), "Shareholders wealth, information signaling, and the specially designated dividend:
an empirical study", Journal of Financial Economics 12:187-209.
Charest, G (1978), "Dividend information, stock returns and market efficiency II", Journal of Financial
Economics 6:297-330.
Chen, N F , B Grundy and R F Stambaugh (1990), "Changing risk, changing risk premiums, and dividend
yield effects", Journal of Business 63:551-570.
Chowdhry, B , and V Nanda (1994), "Repurchase premia as a reason for dividends: a dynamic model
of corporate payout policies", Review of Financial Studies 7:321-350.
Christie, W , and V Nanda (1994), "Free cash flow, shareholder value, and the undistributed profits tax
of 1936 and 1937 ", Journal of Finance 49(5):1727-1754.
Comment, R , and G Jarrell (1991), "The relative power of Dutch-Auction and fixed-priced self-tender
offers and open market share repurchases", Journal of Finance 46(4):1243-1271.
Constantinides, G M (1984), "Optimal stock trading with personal taxes", Journal of Financial Economics
13:65-89.
Constantinides, G M , and B D Grundy (1989), "Optimal investment with stock repurchase and financing
as signals", Review of Financial Studies 2(4):445-466.
Dann, L , R Masulis and D Mayers (1991), "Repurchase tender offers and earning information", Journal
of Accounting and Economics 14:217-251.
De Angelo, H , and L De Angelo (1990), "Dividend policy and financial distress: an empirical investigation
of troubled NYSE firms", Journal of Finance 45 (5):1415-1431.
De Angelo, H , L De Angelo and D Skinner (1996), "Reversal of fortune, dividend signaling and the
disappearance of sustained earnings growth", Journal of Financial Economics 40:341-371.
Del Guercio, D (1996), "The distorting effect of the prudent-man laws on institutional equity
investments", Journal of Financial Economics 40:31-62.
Dubofsky, D A (1992), "A market microstructure explanation of ex-day abnormal returns", Financial
Management (Winter):32-43.
Dunbar, N (2001), "UK firms consider option scheme", Risk Magazine (March), p 14.
Dunsby, A (1993), "Share repurchases and corporate distributions: an empirical study", Working Paper
(University of Pennsylvania).
Eades, K , P Hess and H E Kim (1984), "On interpreting security returns during the ex-dividend period",
Journal of Financial Economics 13:3-34.
Easley, D , S Hvidkjaer and M O'Hara (2002), "Is information risk a determinant of asset returns?",
Journal of Finance 57:2185-2221.
Easterbrook, EH (1984), "Two agency-cost explanations of dividends", American Economic Review
74(4):650-659.
Elton, E , and M Gruber (1970), "Marginal stockholders' tax rates and the clientele effect", Review of
Economics and Statistics 52:68-74.
Fama, E , and K French (2000), "Forecasting profitability and earnings", Journal of Business 73:
161-175.
Fama, E , and K French (2001), "Disappearing dividends: changing firm characteristics or lower
propensity to pay?", Journal of Financial Economics 60:3-43.
Ch 7:
Payout Policy
425
Fama, E E, and H Babiak (1968), "Dividend policy: an empirical analysis", Journal of the American
Statistical Association 63 (324):1132-1161.
Fama, E E, and K R French (1993), "Common risk factors in the returns on stocks and bonds", Journal
of Financial Economics 33:3-56.
Fama, E E, and J D Macbeth (1973), "Risk, return and equilibrium: empirical tests", Journal of Political
Economy 81:607-636.
Federal Reserve Statistical Release (2000), Flow of Funds Accounts of the United States, Federal Reserve
Statistical Release (Board of Governors of the Federal Reserve System, Washington DC).
Fenn, G , and N Liang (2001), "Corporate payout policy and managerial stock incentives", Journal of
Financial Economics 60:45-72.
Fluck, Z (1999), "The dynamics of the manager-shareholder conflict", Review of Financial Studies
12(2):379-404.
Frank, M , and R Jagannathan (1998), "Why do stock prices drop by less than the value of the dividend?
Evidence from a country without taxes", Journal of Financial Economics 47:161-188.
Gonedes, N J (1978), "Corporate signaling, external accounting, and capital market equilibrium: evidence
on dividends, income, and extraordinary items", Journal of Accounting Research 16 (1):26-79.
Gordon, M (1959), "Dividends, earnings and stock prices", Review of Economics and Statistics 41:
99-105.
Green, R , and K Rydqvist (1999), "Ex-day behavior with dividend preference and limitation to shortterm arbitrage: the case of Swedish lottery bonds", Journal of Financial Economics 53:145-187.
Grossman, S J , and O D Hart (1980), "Takeover bids, the free-rider problem, and the theory of the
corporation", Bell Journal of Economics 11:42-54.
Grullon, G , and D Ikenberry (2000), "What do know about stock repurchase?", Journal of Applied
Corporate Finance 13:31-51.
Grullon, G , and R Michaely (2001), "Asymmetric information, agency conflicts and the impact of
taxation on the market reaction to dividend changes", Working Paper (Cornell University, Ithaca, NY).
Grullon, G , and R Michaely (2002), "Dividends, share repurchases and the substitution hypothesis",
The Journal of Finance 62 (4):1649-1684.
Grullon, G , and R Michaely (2003), "The information content of share repurchase programs", The
Journal of Finance, forthcoming.
Grullon, G , R Michaely and B Swaminathan (2002), "Are dividend changes a sign of firm maturity?",
The Journal of Business 75:387-424.
Grullon, G , R Michaely, S Benartzi and R Thaler (2003), "Dividend changes do not signal changes
in future profitability", Working Paper (Cornell University, NY).
Grundy, B (1985), "Trading volume and stock returns around ex-dividend dates", Working Paper
(University of Chicago, Chicago, IL).
Handjinicolaou, G , and A Kalay (1984), "Wealth redistributions or changes in firm value: an analysis
of returns to bondholders and the stockholders around dividend announcements", Journal of Financial
Economics 13 (1):35-63.
Hasbrouck, J , and I Friend (1984), "Why do companies pay dividends?: Comment", American Economic
Review 74:1137-1141.
Healy, PM , and K G Palepu (1988), "Earnings information conveyed by dividend initiations and
omissions", Journal of Financial Economics 21(2):149-176.
Hertzel, M , and P Jain (1991), "Earnings and risk changes around stock repurchase tender offers",
Journal of Accounting and Economics 14:252-274.
Hess, P J (1983), "Test of price effects in the pricing of financial assets", Journal of Business 56:
537-554.
Hubbard, J , and R Michaely (1997), "Do investors ignore dividend taxation? A re-examination of the
Citizen Utilities case", Journal of Financial and Quantitative Analysis 32(1):117-135.
Ikenberry, D , J Lakonishok and T Vermaelen (1995), "Market underreaction to open market share
repurchases", Journal of Financial Economics 39:181-208.
426
F Allen and R Michaely
Ikenberry, D , J Lakonishok and T Vermaelen (2000), "Share repurchases in Canada: performance and
strategic trading", Journal of Finance 55:2373-2397.
Internal Revenue Service (various years), SOI Bulletin (Department of Treasury).
under-valuation,
Jagannathan, M , and C P Stephens (2001), "Motives for open market share repurchases:
Columbia).
Missouri,
of
(University
earnings signaling or free cash flow", Working Paper
and the choice between
Jagannathan, M , C P Stephens and M S Weisbach (2000), "Financial flexibility
dividends and stock repurchases", Journal of Financial Economics 57:355-384.
American
Jensen, M C (1986), "Agency costs of free cash flow, corporate finance, and takeovers",
2):323-329.
(
76
Review
Economic
behavior, agency costs and
Jensen, M C , and WH Meckling (1976), "Theory of the firm: managerial
(
4):305-360.
3
Economics
Financial
of
Journal
structure",
ownership
equilibrium", Journal of
John, K , and J Williams (1985), "Dividends, dilution, and taxes: a signaling
Finance 40(4):1053-1070.
puzzle", Working
Jolls, C (1998), "The role of incentive compensation in explaining the stock repurchase
School).
Paper (Harvard Law
of the clientele
Kalay, A (1982a), "The ex-dividend day behavior of stock prices: a re-examination
1070.
effect", Journal of Finance 37:1059
Journal of Financial
Kalay, A (1982 b), "Stockholder-bondholder conflict and dividend constraint",
Economics 14:423-449.
Financial Management
Kalay, A , and R Michaely (2000), "Dividends and taxes: a reexamination",
29(2):55-75.
from dividend
Kang, S H , and P Kumar (1991), "Determinants of dividend smoothing: Evidence
changes", Working Paper (Carnegie Mellon University, PA).
days: additional
Karpoff, J M , and R A Walkling (1988), "Short-term trading around ex-dividend
2):291-298.
(
21
Economics
Financial
of
Journal
evidence",
stocks", Journal of Financial
Karpoff, J M , and R A Walkling (1990), "Dividend capture in NASDAQ
1/2):39-66.
(
28
Economics
prices: the case of Japan",
Kato, K , and U Loewenstein (1995), "The ex-dividend-day behavior of stock
8:817-847.
Studies
Review of Financial
January returns", Journal
Keim, D (1985), "Dividend yields and stock returns: implications of abnormal
14:473-489.
of Financial Economics
King, M (1977), Public Policy and the Corporation (Chapman and Hall, London).
of trades", Review of
Koski, J, and R Michaely (2000), "Prices, liquidity and the information content
3):659-696.
(
13
Studies
Financial
dividends", Review of
Kumar, P (1988), "Shareholder-manager conflict and the information content of
2):111-136.
1
(
Financial Studies
problems and dividend
La Porta, R , E Lopez-De Silanes, A Shleifer and R Vishny (2000), "Agency
55:1-33.
Finance
of
Journal
world",
policy around the
behavior", Journal of
Lakonishok, J , and T Vermaelen (1983), "Tax reform and the ex-dividend day
Finance 38:1157-1179.
dates", Journal of
Lakonishok, J, and T Vermaelen (1986), "Tax induced trading around ex-dividend
Financial Economics 16:287-319.
flow signaling vs free cash
Lang, L H P , and R H Litzenberger (1989), "Dividend announcements: cash
(
1):181-192.
24
flow hypothesis", Journal of Financial Economics
direct evidence on the
Lewellen, WG , K L Stanley, R C Lease and G G Schlarbaum (1978), "Some
(
5):1385-1399.
33
dividend clientele phenomenon", Journal of Finance
of incremental disbursements",
Lie, E (2000), "Excess funds and the agency problems: an empirical study
Review of Financial Studies 13 (1):219-248.
Paper (College of William
Lie, E (2001), "Financial flexibility and the corporate payout policy", Working
and Mary, VA).
Ch 7:
Payout Policy
427
Lintner, J (1956), "Distribution of incomes of corporations among dividends, retained earnings, and
taxes", American Economic Review 46 (2):97-113.
Litzenberger, R , and K Ramaswamy (1979), "The effects of personal taxes and dividends on capital
asset prices: theory and empirical evidence", Journal of Financial Economics 7:163-195.
Litzenberger, R , and K Ramaswamy (1980), "Dividends, short selling restrictions, tax induced investor
clientele and market equilibrium", Journal of Finance 35:469-482.
Litzenberger, R , and K Ramaswamy (1982), "The effects of dividends on common stock prices: tax
effects or information effects?", Journal of Finance 37:429-443.
Long, J B (1977), "Efficient portfolio choice with differential taxation of dividend and capital gains",
Journal of Financial Economics 5:25-53.
Long Jr, J B (1978), "The market valuation of cash dividends: a case to consider", Journal of Financial
Economics 6 (2/3):235-264.
Loughran, T , and J Ritter (1995), "The new issues puzzle", Journal of Finance 50 (1):23-51.
Lucas, D J , and R L McDonald (1998), "Shareholder heterogeneity, adverse selection, and payout
policy", Journal of Financial and Quantitative Analysis 33 (2):233-253.
Masulis, R W , and A N Korwar (1986), "Seasoned equity offerings: an empirical investigation", Journal
of Financial Economics 15(1/2):91-118.
Michaely, R (1991), "Ex-dividend day stock price behavior: the case of the 1986 Tax Reform Act",
Journal of Finance 46:845-860.
Michaely, R , and M Murgia (1995), "The effect of tax heterogeneity on prices and volume around the
ex-dividend day: evidence from the Milan stock exchange", Review of Financial Studies 8:369-399.
Michaely, R , and W H Shaw (1994), "The pricing of initial public offerings: tests of the adverse selection
and signaling theories", Review of Financial Studies 7(2):279-319.
Michaely, R , and J -L Vila (1995), "Investors' heterogeneity, prices and volume around the ex-dividend
day", Journal of Financial and Quantitative Analysis 30:171-198.
Michaely, R , and J -L Vila (1996), "Trading volume with private valuations: evidence from the exdividend day", Review of Financial Studies 9 (2):471-510.
Michaely, R , R H Thaler and K Womack (1995), "Price reactions to dividend initiations and omissions:
overreaction or drift?", Journal of Finance 50(2):573-608.
Michaely, R , J -L Vila and J Wang (1996), "A model of trading volume with tax-induced heterogeneous
valuation and transaction costs", Journal of Financial Intermediation 5:471-510.
Miller, J , and J McConnell (1995), "Open-market share repurchase programs and bid-ask spreads on
the NYSE: implications for corporate payout policy", Journal of Financial and Quantitative Analysis
30(3):365-382.
Miller, M (1977), "Debt and taxes", The Journal of Finance 32(2):261-275.
Miller, M (1987), "The information content of dividends", in: J Bossons, R Dornbush and S Fischer,
eds , Macroeconomics: Essays in Honor of Franco Modigliani (MIT Press, Cambridge, MA) pp 3761.
Miller, M , and E Modigliani (1961), "Dividend policy, growth and the valuation of shares", Journal of
Business 34:411-433.
Miller, M , and K Rock ( 1985), "Dividend policy under asymmetric information", Journal of Finance
40(4): 1031-1051.
Miller, M , and M Scholes (1978), "Dividends and taxes", Journal of Financial Economics 6:333-264.
Miller, M , and M Scholes (1982), "Dividends and taxes: empirical evidence", Journal of Political
Economy 90:1118-1141.
Moody's dividend record (various years) (Mergert FIS, Inc , New York).
Morgan, I G (1982), "Dividends and capital asset prices", Journal of Finance 37:1071-1086.
Myers, S C (1977), "Determinants of corporate borrowing", Journal of Financial Economics 5(2):
147-175.
Myers, S C (2000), "Outside equity", Journal of Finance 55 (3):1005-1038.
428
E Allen and R Michaely
Myers, S C , and N S Majluf (1984), "Corporate financing and investment decisions when firms have
information that investors do not have", Journal of Financial Economics 13 (2):187-221.
Naranjo, A , M Nimalendran and M Ryngaert (1998), "Stock returns, dividend yield and taxes", Journal
of Finance 53 (6):2029-2057.
Nissim, D , and A Ziv (2001), "Dividend changes and future profitability", Journal of Finance 61 (6):
2111-2134.
Nohel, T , and V Tarhan (1998), "Share repurchases and firm performance: new evidence on the agency
costs of free cash flow", Journal of Financial Economics 49:187-222.
Ofer, A R , and D R Siegel (1987), "Corporate financial policy, information, and market expectations:
an empirical investigation of dividends", Journal of Finance 42 (4):889-911.
Ofer, A R , and A V Thakor (1987), "A theory of stock price responses to alternative corporate cash
disbursement methods: stock repurchases and dividends", Journal of Finance 42 (2):365-394.
Penman, S H (1983), "The predictive content of earnings forecasts and dividends", Journal of Finance
38 (4):1181-1199.
Perez-Gonzalez, F (2000), "Large shareholders and dividends: evidence from U S tax reforms", Working
Paper (Harvard University).
Peterson, P , D Peterson and J Ang (1985), "Direct evidence on the marginal rate of taxation on dividend
income", Journal of Financial Economics 14:267-282.
Pettit, R R (1972), "Dividend announcements, security performance, and capital market efficiency",
Journal of Finance 27 (5):993-1007.
Poterba, J (1986), "The market valuation of cash dividends: the citizens utilities case reconsidered",
Journal of Financial Economics 15:395-406.
Poterba, J , and L H Summers (1984), "New evidence that taxes affect the valuation of dividends",
Journal of Finance 39:1397-1415.
Poterba, J , S Venti and D Wise (1996), "How retirement saving programs increase saving Reconciling
the evidence", Journal of Economic Perspectives 10 (4):91-112.
Richardson, G , S Sefcik and R Thompson (1986), "A test of dividend irrelevance using volume
reactions to a change in dividend policy", Journal of Financial Economics 17:313-333.
Ritter, J R (1991), "The long run performance of initial public offerings", Journal of Finance 46 (1):3-28.
Rosenberg, B , and V Marathe (1979), "Tests of capital asset pricing hypotheses", Research in Finance
1:115-224.
Ross, S A (1977), "The determination of financial structure: the incentive signalling approach", Bell
Journal of Economics, pp 13-40.
Sarig, O (1984), "Why do companies pay dividends?: Comment", American Economic Review 74:1142.
Shefrin, H M , and M Statman (1984), "Explaining investor preference for cash dividends", Journal of
Financial Economics 13(2):253-282.
Shleifer, A , and R Vishny (1986), "Large shareholders and corporate control", Journal of Political
Economy 94(3):461-488.
Standard and Poor's Dividend Record (various years) (Standard and Poor's Corporation, New York).
Stephens, C , and M Weisbach (1998), "Actual share reacquisitions in open market repurchases
programs", Journal of Finance 53(1):313-333.
Stiglitz, J E (1983), "Some aspects of the taxation of capital gains", Journal of Public Economics
21:257-296.
Stulz, R (1988), "Managerial control of voting rights: financing policies and the market for corporate
control", Journal of Financial Economics 20(1-2):25-54.
Thaler, R H , and H M Shefrin (1981), "An economic theory of self-control", Journal of Political
Economy 89(2):392-406.
Vermaelen, T (1981), "Common stock repurchases and market signaling: an empirical study", Journal
of Financial Economics 9(2):138-183.
Vermaelen, T (1984), "Repurchase tender offers, signalling and managerial incentives", Journal of
Financial and Quantitative Analysis 19:163-181.
Ch 7:
Payout Policy
429
Watts, R (1973), "The information content of dividends", Journal of Business 46 (2):191-211.
Weisbenner, S (2000), "Corporate share repurchase in the mid-1990 s: What role do stock options play",
Working Paper (MIT, MA).
Williams, J (1988), "Efficient signaling with dividends, investment, and stock repurchases", Journal of
Finance 43(3):737-747.
Yoon, P S , and L Starks (1995), "Signaling, investment opportunities, and dividend announcements",
Review of Financial Studies 8(4):995-1018.
Zwiebel, J (1996), "Dynamic capital structure under managerial entrenchment", American Economic
Review 86(5):1197-1215.